GPT-6 Astra
Unavailable in this round
Head-to-head comparison
GPT-6 Astra vs. DeepSeek-V4-Flash
Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.
- shared result series
- 62
- distinct benchmark metrics
- 2
- latest retrieval
- 2026-10-15
Change model selection
Specifications
General specifications
Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.
| Attribute | GPT-6 Astra | DeepSeek-V4-Flash |
|---|---|---|
| Provider | OpenAI | DeepSeek |
| Released | Sep 3, 2026 | Apr 24, 2026 |
| Availability | Limited access | Limited access |
| Model type | Proprietary | Open Weights |
| Parameters | Not published | 284B, 13B active |
| Architecture | Not published | Mixture of Experts |
| Context window | 1,050,000 tokens | 1,000,000 tokens |
| Knowledge cutoff | April 30, 2026 | Not published |
| Technical sources | Model, Context | Model |
Pricing
API pricing compared
The table compares official standard API rates per 1M tokens, including published cache pricing. Batch, fast, priority, regional surcharges, tool calls, and cloud platform pricing are excluded.
| Price type | GPT-6 Astra | DeepSeek-V4-Flash |
|---|---|---|
| API input | $10 ≤272K / $20 >272K | $0.13 |
| API output | $50 ≤272K / $75 >272K | $0.28 |
| Cache read | $1 ≤272K / $2 >272K | Not listed |
| Cache write (5 min.) | $12.5 ≤272K / $25 >272K | Not listed |
| Cache write (1 hr.) | Not listed | Not listed |
| Price verified | 09/22/2026 Source | 10/19/2026 Source |
GPT-6 AstraDeepSeek-V4-Flash
API inputUSD per 1M tokens, base tier
GPT-6 Astra$10
DeepSeek-V4-FlashLower price$0.13
API outputUSD per 1M tokens, base tier
GPT-6 Astra$50
DeepSeek-V4-FlashLower price$0.28
Key to the marks:better valuebetter, but not conclusivetie
Pricing by provider
Direct developer pricing and provider endpoint tariffs routed through OpenRouter are listed separately. Prices are in USD per 1M tokens. Context-dependent direct tiers plus regional, Flex, Priority, and other OpenRouter tariffs remain individually identifiable.
| Model | Provider | Purchase route and tariff | Input per 1M tokens | Output per 1M tokens | Source |
|---|---|---|---|---|---|
| GPT-6 Astra | OpenAI | Direct from developerUp to 272,000 context tokens | $10 | $50 | Source09/22/2026 |
| OpenAI | Direct from developerAbove 272,000 context tokens | $20 | $75 | Source09/22/2026 | |
| Amazon Bedrock | Via OpenRouteramazon-bedrock/us-west-2 | $11 | $55 | Source10/19/2026 | |
| Azure | Via OpenRouterazure | $10 | $50 | Source10/19/2026 | |
| Azure | Via OpenRouterazure/us | $11 | $55 | Source10/19/2026 | |
| OpenAI | Via OpenRouteropenai | $10 | $50 | Source10/19/2026 | |
| OpenAI | Via OpenRouteropenai/fast | $20 | $100 | Source10/19/2026 | |
| OpenAI | Via OpenRouteropenai/flex | Lowest listed price: $5 | Lowest listed price: $25 | Source10/19/2026 | |
| OpenAI | Via OpenRouteropenai/ultrafast | $60 | $300 | Source10/19/2026 | |
| DeepSeek-V4-Flash | Alibaba | Via OpenRouteralibaba | $0.176 | $0.528 | Source10/19/2026 |
| Alibaba | Via OpenRouteralibaba/us-east-1 | $0.187 | $0.561 | Source10/19/2026 | |
| AtlasCloud | Via OpenRouteratlas-cloud/fp4 | $0.44 | $1.32 | Source10/19/2026 | |
| Baidu | Via OpenRouterbaidu/fp8 | $0.0598 | $0.1795 | Source10/19/2026 | |
| BaseTen | Via OpenRouterbaseten/fp32 | $0.13 | $0.26 | Source10/19/2026 | |
| BaseTen | Via OpenRouterbaseten/fp8 | $0.13 | $0.26 | Source10/19/2026 | |
| Cloudflare | Via OpenRoutercloudflare | $0.44 | $1.32 | Source10/19/2026 | |
| Cohere | Via OpenRoutercohere | $0.14 | $0.28 | Source10/19/2026 | |
| CoreWeave | Via OpenRoutercoreweave/fp8 | $0.13 | $0.28 | Source10/19/2026 | |
| DeepInfra | Via OpenRouterdeepinfra/fp8 | $0.06 | $0.18 | Source10/19/2026 | |
| DigitalOcean | Via OpenRouterdigitalocean | $0.119 | $0.238 | Source10/19/2026 | |
| GMICloud | Via OpenRoutergmicloud/fp8 | $0.286 | $0.858 | Source10/19/2026 | |
| Inceptron | Via OpenRouterinceptron/fp4 | $0.13 | $0.3996 | Source10/19/2026 | |
| Mancer 2 | Via OpenRoutermancer/fp8 | $0.2 | $0.6 | Source10/19/2026 | |
| Novita | Via OpenRouternovita/fp8 | $0.4092 | $1.2276 | Source10/19/2026 | |
| OpenInference | Via OpenRouteropen-inference/fp4 | Lowest listed price: $0.0052 | Lowest listed price: $0.12 | Source10/19/2026 | |
| Parasail | Via OpenRouterparasail/fp8 | $0.14 | $0.28 | Source10/19/2026 | |
| Phala | Via OpenRouterphala | $0.308 | $0.924 | Source10/19/2026 | |
| Reka | Via OpenRouterreka | $0.021 | $0.528 | Source10/19/2026 | |
| Relace | Via OpenRouterrelace/fp4 | $0.0065 | $0.45 | Source10/19/2026 | |
| Sail Research | Via OpenRoutersail-research/fp4 | $0.15 | $0.15 | Source10/19/2026 | |
| Sail Research | Via OpenRoutersail-research/us | $0.15 | $0.15 | Source10/19/2026 | |
| SiliconFlow | Via OpenRoutersiliconflow/fp8 | $0.22 | $0.66 | Source10/19/2026 | |
| StreamLake | Via OpenRouterstreamlake/fp8 | $0.088 | $0.264 | Source10/19/2026 | |
| Together | Via OpenRoutertogether | $0.14 | $0.28 | Source10/19/2026 | |
| Venice | Via OpenRoutervenice | $0.13 | $0.26 | Source10/19/2026 | |
| Wafer | Via OpenRouterwafer/fast | $0.13 | $0.23 | Source10/19/2026 |
Performance
Shared benchmarks
Only published results with the same benchmark version, task, metric, and comparison cohort are paired. Different reasoning levels, agents, harnesses, or output limits appear directly in the table.
Capability profile from matched benchmarks
Each axis averages directly matched benchmark families. A value of 100 means the stronger value within this pair, not a universal quality score.
Not enough directly matched categories for a reliable radar
Only 0 of at least 5 categories have comparable, direct model measurements. The benchmark tables below show the available evidence without filling missing axes.
GPT-6 AstraDeepSeek-V4-Flash
LMArena Code Arena WebDevOverall. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,780.244
DeepSeek-V4-Flash1,430.632
LMArena Text, Style ControlOverall. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,476.716
DeepSeek-V4-Flash1,435.976
LMArena Code Arena WebDevWebdev. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,780.244
DeepSeek-V4-Flash1,430.632
LMArena Text, Style ControlOverall. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,476.716
DeepSeek-V4-Flash1,438.654
LMArena Code Arena WebDevWebdev-html. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,780.13
DeepSeek-V4-Flash1,416.219
LMArena Text, Style ControlBusiness, management and finance. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,472.34
DeepSeek-V4-Flash1,438.829
LMArena Code Arena WebDevWebdev-react. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,804.405
DeepSeek-V4-Flash1,421.593
LMArena Text, Style ControlBusiness, management and finance. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,472.34
DeepSeek-V4-Flash1,434.084
LMArena Text, Style ControlChinese. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,514.537
DeepSeek-V4-Flash1,470.018
LMArena Text, Style ControlChinese. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,514.537
DeepSeek-V4-Flash1,480.516
LMArena Text, Style ControlCoding. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,543.161
DeepSeek-V4-Flash1,484.301
LMArena Text, Style ControlCoding. Higher is better. Test setup not documented as identical
GPT-6 AstraHigher measured value1,543.161
DeepSeek-V4-Flash1,479.463
Key to the marks:better valuebetter, but not conclusivetie
| Benchmark and task | GPT-6 Astra | DeepSeek-V4-Flash | Source |
|---|---|---|---|
LMArena Code Arena WebDevOverallTest detailsTest setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Code Arena WebDev DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Code Arena WebDevGPT-6 Astra: Arena rating from 5,746 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 10,152 published pairwise votes. | Higher measured value1,780.244 95% confidence interval: ±10.541The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,430.632 95% confidence interval: ±6.942The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlOverallTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 11,674 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 52,016 published pairwise votes. | Higher measured value1,476.716 95% confidence interval: ±6.385The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,435.976 95% confidence interval: ±4.036The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Code Arena WebDevWebdevTest detailsTest setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Code Arena WebDev DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Code Arena WebDevGPT-6 Astra: Arena rating from 5,746 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 10,152 published pairwise votes. | Higher measured value1,780.244 95% confidence interval: ±10.541The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,430.632 95% confidence interval: ±6.942The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlOverallTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 11,674 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 51,814 published pairwise votes. | Higher measured value1,476.716 95% confidence interval: ±6.385The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,438.654 95% confidence interval: ±4.106The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Code Arena WebDevWebdev-htmlTest detailsTest setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Code Arena WebDev DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Code Arena WebDevGPT-6 Astra: Arena rating from 1,129 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 1,165 published pairwise votes. | Higher measured value1,780.13 95% confidence interval: ±22.887The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,416.219 95% confidence interval: ±18.033The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlBusiness, management and financeTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,061 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 10,448 published pairwise votes. | Higher measured value1,472.34 95% confidence interval: ±13.336The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,438.829 95% confidence interval: ±6.961The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Code Arena WebDevWebdev-reactTest detailsTest setup not documented as identicalNo direct winnerVersion: webdevMetric: WebDev Arena Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Code Arena WebDev DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Code Arena WebDevGPT-6 Astra: Arena rating from 3,422 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 8,109 published pairwise votes. | Higher measured value1,804.405 95% confidence interval: ±13.548The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,421.593 95% confidence interval: ±7.665The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlBusiness, management and financeTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,061 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 10,643 published pairwise votes. | Higher measured value1,472.34 95% confidence interval: ±13.336The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,434.084 95% confidence interval: ±6.963The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlChineseTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 877 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 3,005 published pairwise votes. | Higher measured value1,514.537 95% confidence interval: ±19.541The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,470.018 95% confidence interval: ±11.382The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlChineseTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 877 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 2,975 published pairwise votes. | Higher measured value1,514.537 95% confidence interval: ±19.541The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,480.516 95% confidence interval: ±11.412The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlCodingTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,659 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 15,290 published pairwise votes. | Higher measured value1,543.161 95% confidence interval: ±11.756The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,484.301 95% confidence interval: ±6.214The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlCodingTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,659 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 15,245 published pairwise votes. | Higher measured value1,543.161 95% confidence interval: ±11.756The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,479.463 95% confidence interval: ±6.25The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlCreative writingTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,620 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 8,595 published pairwise votes. | Higher measured value1,449.578 95% confidence interval: ±12.436The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,408.246 95% confidence interval: ±7.512The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlCreative writingTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,620 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 8,668 published pairwise votes. | Higher measured value1,449.578 95% confidence interval: ±12.436The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,410.845 95% confidence interval: ±7.548The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlEnglishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 4,703 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 23,136 published pairwise votes. | Higher measured value1,485.113 95% confidence interval: ±9.08The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,445.189 95% confidence interval: ±5.273The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlEnglishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 4,703 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 23,022 published pairwise votes. | Higher measured value1,485.113 95% confidence interval: ±9.08The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,449.707 95% confidence interval: ±5.31The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlEntertainment, sports and mediaTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,996 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 11,221 published pairwise votes. | Higher measured value1,458.633 95% confidence interval: ±11.598The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,404.55 95% confidence interval: ±6.918The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlEntertainment, sports and mediaTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,996 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 11,460 published pairwise votes. | Higher measured value1,458.633 95% confidence interval: ±11.598The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,407.781 95% confidence interval: ±6.873The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlExcluding tiesTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 8,636 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 39,351 published pairwise votes. | Higher measured value1,488.868 95% confidence interval: ±8.374The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,431.858 95% confidence interval: ±5.294The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlExcluding tiesTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 8,636 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 39,449 published pairwise votes. | Higher measured value1,488.868 95% confidence interval: ±8.374The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,436.129 95% confidence interval: ±5.336The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlExpert promptsTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,132 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 5,416 published pairwise votes. | Higher measured value1,509.902 95% confidence interval: ±17.672The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,457.085 95% confidence interval: ±8.727The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlExpert promptsTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,132 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 5,338 published pairwise votes. | Higher measured value1,509.902 95% confidence interval: ±17.672The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,466.764 95% confidence interval: ±8.816The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlFrenchTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 333 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 2,004 published pairwise votes. | Higher measured value1,489.442 95% confidence interval: ±32.984The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,449.161 95% confidence interval: ±14.911The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlFrenchTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 333 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 1,957 published pairwise votes. | Higher measured value1,489.442 95% confidence interval: ±32.984The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,458.886 95% confidence interval: ±15.182The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlGermanTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 176 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 841 published pairwise votes. | Higher measured value1,480.596 95% confidence interval: ±44.81The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,425.17 95% confidence interval: ±21.221The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlGermanTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 176 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 904 published pairwise votes. | Higher measured value1,480.596 95% confidence interval: ±44.81The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,433.291 95% confidence interval: ±20.525The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlHard promptsTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 7,117 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 34,789 published pairwise votes. | Higher measured value1,502.017 95% confidence interval: ±7.647The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,459.087 95% confidence interval: ±4.755The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlHard promptsTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 7,117 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 34,719 published pairwise votes. | Higher measured value1,502.017 95% confidence interval: ±7.647The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,458.022 95% confidence interval: ±4.823The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlHard prompts, EnglishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,847 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 16,226 published pairwise votes. | Higher measured value1,505.128 95% confidence interval: ±11.467The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,463.782 95% confidence interval: ±6.043The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlHard prompts, EnglishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 2,847 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 16,101 published pairwise votes. | Higher measured value1,505.128 95% confidence interval: ±11.467The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,464.326 95% confidence interval: ±6.118The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlInstruction followingTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 3,666 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 17,984 published pairwise votes. | Higher measured value1,471.988 95% confidence interval: ±10.084The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,428.161 95% confidence interval: ±5.795The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlInstruction followingTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 3,666 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 18,097 published pairwise votes. | Higher measured value1,471.988 95% confidence interval: ±10.084The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,433.654 95% confidence interval: ±5.864The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlJapaneseTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 220 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 653 published pairwise votes. | Higher measured value1,429.026 95% confidence interval: ±42.875The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,398.855 95% confidence interval: ±24.754The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlJapaneseTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 220 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 640 published pairwise votes. | Higher measured value1,429.026 95% confidence interval: ±42.875The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,426.069 95% confidence interval: ±24.977The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlKoreanTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 210 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 900 published pairwise votes. | Higher measured value1,450.643 95% confidence interval: ±41.535The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,391.938 95% confidence interval: ±20.897The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlKoreanTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 210 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 949 published pairwise votes. | Higher measured value1,450.643 95% confidence interval: ±41.535The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,376.332 95% confidence interval: ±20.813The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlLegal and governmentTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,000 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 4,152 published pairwise votes. | Higher measured value1,486.122 95% confidence interval: ±18.889The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,448.281 95% confidence interval: ±9.836The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlLegal and governmentTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,000 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 4,099 published pairwise votes. | Higher measured value1,486.122 95% confidence interval: ±18.889The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,457.757 95% confidence interval: ±9.897The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlLife, physical and social scienceTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,987 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 8,416 published pairwise votes. | Higher measured value1,489.3 95% confidence interval: ±13.614The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,456.243 95% confidence interval: ±7.368The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlLife, physical and social scienceTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,987 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 8,489 published pairwise votes. | Higher measured value1,489.3 95% confidence interval: ±13.614The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,459.964 95% confidence interval: ±7.4The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlLonger queriesTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 4,917 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 23,382 published pairwise votes. | Higher measured value1,488.357 95% confidence interval: ±9.148The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,447.798 95% confidence interval: ±5.56The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlLonger queriesTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 4,917 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 23,493 published pairwise votes. | Higher measured value1,488.357 95% confidence interval: ±9.148The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,448.579 95% confidence interval: ±5.563The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMathTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 587 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 2,699 published pairwise votes. | Higher measured value1,481.638 95% confidence interval: ±23.784The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,424.989 95% confidence interval: ±11.731The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMathTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 587 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 2,625 published pairwise votes. | Higher measured value1,481.638 95% confidence interval: ±23.784The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,441.363 95% confidence interval: ±11.858The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMathematical professionsTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 664 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 2,980 published pairwise votes. | Higher measured value1,495.728 95% confidence interval: ±22.879The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,434.613 95% confidence interval: ±11.415The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMathematical professionsTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 664 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 2,825 published pairwise votes. | Higher measured value1,495.728 95% confidence interval: ±22.879The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,444.604 95% confidence interval: ±11.746The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMedicine and healthcareTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 841 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 3,692 published pairwise votes. | Higher measured value1,469.187 95% confidence interval: ±20.767The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,459.398 95% confidence interval: ±10.484The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMedicine and healthcareTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 841 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 3,760 published pairwise votes. | Higher measured value1,469.187 95% confidence interval: ±20.767The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,461.041 95% confidence interval: ±10.433The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMulti-turnTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,593 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 9,349 published pairwise votes. | Higher measured value1,485.974 95% confidence interval: ±15.121The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,453.168 95% confidence interval: ±7.275The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlMulti-turnTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,593 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 9,223 published pairwise votes. | Higher measured value1,485.974 95% confidence interval: ±15.121The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,446.213 95% confidence interval: ±7.326The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlNon-EnglishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 6,968 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 28,879 published pairwise votes. | Higher measured value1,463.829 95% confidence interval: ±7.687The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,422.96 95% confidence interval: ±4.937The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlNon-EnglishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 6,968 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 28,790 published pairwise votes. | Higher measured value1,463.829 95% confidence interval: ±7.687The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,423.369 95% confidence interval: ±5.014The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlPolishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 159 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 1,027 published pairwise votes. | Higher measured value1,454.351 95% confidence interval: ±45.691The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,436.594 95% confidence interval: ±18.719The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlPolishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 159 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 1,040 published pairwise votes. | Higher measured value1,454.351 95% confidence interval: ±45.691The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,433.769 95% confidence interval: ±18.399The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlRussianTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,542 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 5,766 published pairwise votes. | Higher measured value1,469.089 95% confidence interval: ±15.197The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,432.997 95% confidence interval: ±8.479The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlRussianTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 1,542 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 5,793 published pairwise votes. | Higher measured value1,469.089 95% confidence interval: ±15.197The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,436.964 95% confidence interval: ±8.567The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlSoftware and IT servicesTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 4,224 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 21,384 published pairwise votes. | Higher measured value1,522.269 95% confidence interval: ±9.455The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,471.26 95% confidence interval: ±5.499The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlSoftware and IT servicesTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 4,224 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 21,349 published pairwise votes. | Higher measured value1,522.269 95% confidence interval: ±9.455The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,472.791 95% confidence interval: ±5.525The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlSpanishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 345 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 1,580 published pairwise votes. | Higher measured value1,443.13 95% confidence interval: ±31.667The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,431.488 95% confidence interval: ±16.062The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlSpanishTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 345 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 1,476 published pairwise votes. | Higher measured value1,443.13 95% confidence interval: ±31.667The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,429.907 95% confidence interval: ±16.461The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlWriting, literature and languageTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 3,276 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 12,619 published pairwise votes. | Higher measured value1,464.251 95% confidence interval: ±11.022The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,415.784 95% confidence interval: ±6.537The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
LMArena Text, Style ControlWriting, literature and languageTest detailsTest setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-6 Astra: Snapshot: gpt-6-astra-max, Reasoning: max, Harness: Arena Text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena Text, style controlGPT-6 Astra: Arena rating from 3,276 published pairwise votes.DeepSeek-V4-Flash: Arena rating from 12,789 published pairwise votes. | Higher measured value1,464.251 95% confidence interval: ±11.022The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | 1,422.178 95% confidence interval: ±6.53The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result. | LMArenaSource detailsData as of: 10/13/2026 |
Key to the marks:better valuebetter, but not conclusivetie | |||
Our model tests
Our tests
SVG short test
One SVG prompt and fixed output settings. Captures are shown only with 0 reported reasoning tokens or documented lack of reasoning controls. Identical test conditions do not make the models' internal compute identical.
Shared test round: 4ae5d8eeb9a9
DeepSeek-V4-Flash
API request failed. The model was not evaluated.
Earlier tests
Pelican on a bicycle
A community classic for free-form SVG drawing.
GPT-6 Astra
Reasoning (requested): high
Current attempt
- Run date
- Sep 7, 2026
- API provider
- openrouter
- Requested API model
- openai/gpt-6-astra
- API model reported in response
- openai/gpt-6-astra
- Reasoning setting
- high
- Pinned endpoint
- OpenAI | openai/gpt-6-astra-20260903
- Temperature
- Provider default
- Token limit including reasoning
- 8,192 tokens
- Test protocol
- community-visual-tests-v1
- Technical checks
- Passed, not a quality rating
- Response time
- 141.7 s
Visually reviewed. This individual test does not establish overall model quality.
DeepSeek-V4-Flash
The response reached the token limit. No complete recording is available.
Reasoning (requested): high
Current attempt
- Run date
- Oct 5, 2026
- API provider
- openrouter
- Requested API model
- deepseek/deepseek-v4-flash-0731
- API model reported in response
- deepseek/deepseek-v4-flash-0731
- Reasoning setting
- high
- Pinned endpoint
- StreamLake | deepseek/deepseek-v4-flash-20260731
- Temperature
- Provider default
- Token limit including reasoning
- 8,192 tokens
- Test protocol
- community-visual-tests-v1
- Technical checks
- Failed
- Response time
- 97.8 s
Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.
Earlier attempts (3)
- Oct 2, 2026API request failedReasoning setting: high
- Oct 1, 2026API request failedReasoning setting: high
- Sep 7, 2026API request failedReasoning setting: high
Prompt and test conditions
Generate an SVG of a pelican riding a bicycle
Token limit including reasoning: 8,192 tokens
Task origin (Simon Willison)How to read this comparison
Winning one benchmark is not an overall verdict
Your workload, budget, and required context length matter most. A coding benchmark says little about visual understanding or agent performance.
The standard price table uses direct developer rates. The provider comparison labels OpenRouter endpoints, regions, and special tariffs separately.
Every result links to its measurement source and retrieval date. Labels distinguish vendor reports, official benchmark leaderboards, and independent evaluations. Disputed or archived results are excluded.
More matchups
Related LLM comparisons
Compare GPT-6 Astra and DeepSeek-V4-Flash with other leading models using the same data and benchmark logic.