Head-to-head comparison

Gemini 3.5 Flash 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: 31
distinct benchmarks: 2
latest retrieval: 2026-09-29

First modelSecond model

Specifications

General specifications

Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.

Specifications for Gemini 3.5 Flash and DeepSeek-V4-Flash
Attribute	Gemini 3.5 Flash	DeepSeek-V4-Flash
Provider	Google	DeepSeek
Released	May 19, 2026	Apr 24, 2026
Availability	Active	Limited access
Model type	Proprietary	Open Weights
Parameters	Not published	284B, 13B active
Architecture	Not published	Mixture of Experts
Context window	1,048,576 tokens	1,000,000 tokens
Knowledge cutoff	Not published	Not published
Technical sources	Model	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.

API pricing for Gemini 3.5 Flash and DeepSeek-V4-Flash
Price type	Gemini 3.5 Flash	DeepSeek-V4-Flash
API input	$1.5	$0.14
API output	$9	$0.4
Cache read	$0.15	Not listed
Cache write (5 min.)	Not listed	Not listed
Cache write (1 hr.)	Not listed	Not listed
Price verified	09/22/2026 Source	10/02/2026 Source

Gemini 3.5 FlashDeepSeek-V4-Flash

API inputUSD per 1M tokens, base tier

Gemini 3.5 Flash$1.5

DeepSeek-V4-Flash$0.142

API outputUSD per 1M tokens, base tier

Gemini 3.5 Flash$9

DeepSeek-V4-Flash$0.4

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.

Provider pricing for Gemini 3.5 Flash and DeepSeek-V4-Flash
Model	Provider	Purchase route and tariff	Input per 1M tokens	Output per 1M tokens	Source
Gemini 3.5 Flash	Google	Direct from developerStandard API	$1.5	$9	Source09/22/2026
	Google	Via OpenRoutergoogle-vertex/global	$1.5	$9	Source10/02/2026
	Google	Via OpenRoutergoogle-vertex/global/flex	Lowest listed price: $0.75	Lowest listed price: $4.5	Source10/02/2026
	Google	Via OpenRoutergoogle-vertex/global/priority	$2.7	$16.2	Source10/02/2026
	Google	Via OpenRoutergoogle-vertex/us	$1.65	$9.9	Source10/02/2026
	Google AI Studio	Via OpenRoutergoogle-ai-studio	$1.5	$9	Source10/02/2026
	Google AI Studio	Via OpenRoutergoogle-ai-studio/flex	Lowest listed price: $0.75	Lowest listed price: $4.5	Source10/02/2026
	Google AI Studio	Via OpenRoutergoogle-ai-studio/priority	$2.7	$16.2	Source10/02/2026
DeepSeek-V4-Flash	Alibaba	Via OpenRouteralibaba	$0.176	$0.528	Source10/02/2026
	AtlasCloud	Via OpenRouteratlas-cloud/fp4	$0.44	$1.32	Source10/02/2026
	Baidu	Via OpenRouterbaidu/fp8	$0.44	$1.32	Source10/02/2026
	BaseTen	Via OpenRouterbaseten/fp8#1	$0.13	$0.26	Source10/02/2026
	BaseTen	Via OpenRouterbaseten/fp8#2	$0.13	$0.26	Source10/02/2026
	Cloudflare	Via OpenRoutercloudflare	$0.44	$1.32	Source10/02/2026
	Cohere	Via OpenRoutercohere	$0.14	$0.28	Source10/02/2026
	CoreWeave	Via OpenRoutercoreweave/fp8	$0.13	$0.28	Source10/02/2026
	DeepInfra	Via OpenRouterdeepinfra/fp8	$0.06	$0.18	Source10/02/2026
	DigitalOcean	Via OpenRouterdigitalocean	$0.119	$0.238	Source10/02/2026
	GMICloud	Via OpenRoutergmicloud/fp8	$0.286	$0.858	Source10/02/2026
	Inceptron	Via OpenRouterinceptron/fp4	$0.05	$0.65	Source10/02/2026
	Makora	Via OpenRoutermakora	$0.09	$0.195	Source10/02/2026
	Mancer 2	Via OpenRoutermancer/fp8	$0.2	$0.6	Source10/02/2026
	Morph	Via OpenRoutermorph/bf16	$0.142	$0.4	Source10/02/2026
	NextBit	Via OpenRouternextbit/fp8	$0.352	$1.056	Source10/02/2026
	Novita	Via OpenRouternovita/fp8	$0.4092	$1.2276	Source10/02/2026
	OpenInference	Via OpenRouteropen-inference/fp8	Lowest listed price: $0.0038	$1.6	Source10/02/2026
	Parasail	Via OpenRouterparasail/fp8	$0.14	$0.28	Source10/02/2026
	Phala	Via OpenRouterphala	$0.308	$0.924	Source10/02/2026
	Reka	Via OpenRouterreka	$0.021	$0.528	Source10/02/2026
	Relace	Via OpenRouterrelace/fp4	$0.0051	$1.28	Source10/02/2026
	Sail Research	Via OpenRoutersail-research/fp4	$0.019	$0.3	Source10/02/2026
	Sail Research	Via OpenRoutersail-research/us	$0.019	$0.42	Source10/02/2026
	SiliconFlow	Via OpenRoutersiliconflow/fp8	$0.22	$0.66	Source10/02/2026
	StreamLake	Via OpenRouterstreamlake/fp8	$0.044	Lowest listed price: $0.132	Source10/02/2026
	Together	Via OpenRoutertogether	$0.14	$0.28	Source10/02/2026
	Venice	Via OpenRoutervenice	$0.175	$0.35	Source10/02/2026
	Wafer	Via OpenRouterwafer/fast	$0.12	$0.7	Source10/02/2026

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.

Gemini 3.5 Flash

DeepSeek-V4-Flash

Coding: 100 / 98.2 (1 benchmark family)

Reasoning: 100 / 98 (1 benchmark family)

Mathematics: 100 / 96.9 (1 benchmark family)

Knowledge: 100 / 97.8 (1 benchmark family)

Finance: 100 / 97.7 (1 benchmark family)

Legal: 100 / 98.7 (1 benchmark family)

Cybersecurity: 97.2 / 94.7 (1 benchmark family)

Health: 100 / 98.5 (1 benchmark family)

31 matched rows from 2 benchmark families were considered. The radar shows 8 of 9 comparable categories.

Gemini 3.5 FlashDeepSeek-V4-Flash

CyberBench v1.1Patch. Higher is better. No documented setup difference

Gemini 3.5 Flash81.034 %

DeepSeek-V4-Flash72.414 %

LMArena Text, Style ControlOverall. Higher is better. No documented setup difference

Gemini 3.5 Flash1,476.958

DeepSeek-V4-Flash1,438.511

CyberBench v1.1PoC. Higher is better. No documented setup difference

Gemini 3.5 Flash57.627 %

DeepSeek-V4-Flash61.017 %

LMArena Text, Style ControlBusiness, management and finance. Higher is better. No documented setup difference

Gemini 3.5 Flash1,467.855

DeepSeek-V4-Flash1,434.189

LMArena Text, Style ControlChinese. Higher is better. No documented setup difference

Gemini 3.5 Flash1,514.086

DeepSeek-V4-Flash1,479.924

LMArena Text, Style ControlCoding. Higher is better. No documented setup difference

Gemini 3.5 Flash1,507.378

DeepSeek-V4-Flash1,479.555

LMArena Text, Style ControlCreative writing. Higher is better. No documented setup difference

Gemini 3.5 Flash1,462.846

DeepSeek-V4-Flash1,410.703

LMArena Text, Style ControlEnglish. Higher is better. No documented setup difference

Gemini 3.5 Flash1,478.707

DeepSeek-V4-Flash1,449.94

LMArena Text, Style ControlEntertainment, sports and media. Higher is better. No documented setup difference

Gemini 3.5 Flash1,452.16

DeepSeek-V4-Flash1,408.063

LMArena Text, Style ControlExcluding ties. Higher is better. No documented setup difference

Gemini 3.5 Flash1,486.83

DeepSeek-V4-Flash1,436.008

LMArena Text, Style ControlExpert prompts. Higher is better. No documented setup difference

Gemini 3.5 Flash1,490.882

DeepSeek-V4-Flash1,466.306

LMArena Text, Style ControlFrench. Higher is better. No documented setup difference

Gemini 3.5 Flash1,494.334

DeepSeek-V4-Flash1,459.263

Key to the marks:better valuebetter, but not conclusivetie

Our model tests

Our tests

Compare how the models respond to the same prompt. Each test shows the first attempt, with no subsequent fixes to the generated code. These results do not contribute to an overall score.

Pelican on a bicycle

A community classic for free-form SVG drawing.

Gemini 3.5 Flash

API access was unavailable for this run. The model was not evaluated.

Reasoning (requested): high

First attempt

Run date: 7 Sept 2026
API provider: openrouter
Requested API model: google/gemini-3.5-flash
Reasoning setting: high
Pinned endpoint: Google AI Studio | google/gemini-3.5-flash-20260519
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Not run

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

DeepSeek-V4-Flash

API access was unavailable for this run. The model was not evaluated.

Reasoning (requested): high

First attempt

Run date: 7 Sept 2026
API provider: openrouter
Requested API model: deepseek/deepseek-v4-flash-0731
Reasoning setting: high
Pinned endpoint: DeepSeek | deepseek/deepseek-v4-flash-20260731
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Not run

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

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

Compare Gemini 3.5 Flash and DeepSeek-V4-Flash with other leading models using the same data and benchmark logic.

View the complete LLM comparison

Benchmark scores for Gemini 3.5 Flash and DeepSeek-V4-Flash
Benchmark and task	Gemini 3.5 Flash	DeepSeek-V4-Flash	Source
CyberBench v1.1Patch Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner81.034 %	72.414 %	Vals AI Source details Data as of: 08/03/2026
LMArena Text, Style ControlOverall Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 46,190 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 51,747 anonymous pairwise votes.	Winner1,476.958 95% confidence interval: ±3.954The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,438.511 95% confidence interval: ±4.115The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
CyberBench v1.1PoC Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	57.627 %	Winner61.017 %	Vals AI Source details Data as of: 08/03/2026
LMArena Text, Style ControlBusiness, management and finance Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 9,088 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 10,641 anonymous pairwise votes.	Winner1,467.855 95% confidence interval: ±7.031The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,434.189 95% confidence interval: ±6.982The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlChinese Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 3,491 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 2,992 anonymous pairwise votes.	Winner1,514.086 95% confidence interval: ±10.738The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,479.924 95% confidence interval: ±11.42The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlCoding Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 13,289 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 15,239 anonymous pairwise votes.	Winner1,507.378 95% confidence interval: ±6.21The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,479.555 95% confidence interval: ±6.265The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlCreative writing Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 8,808 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 8,638 anonymous pairwise votes.	Winner1,462.846 95% confidence interval: ±7.404The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,410.703 95% confidence interval: ±7.575The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlEnglish Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 19,425 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 22,908 anonymous pairwise votes.	Winner1,478.707 95% confidence interval: ±5.278The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,449.94 95% confidence interval: ±5.323The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlEntertainment, sports and media Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 11,195 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 11,428 anonymous pairwise votes.	Winner1,452.16 95% confidence interval: ±6.734The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,408.063 95% confidence interval: ±6.896The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlExcluding ties Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 34,820 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 39,391 anonymous pairwise votes.	Winner1,486.83 95% confidence interval: ±5.234The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,436.008 95% confidence interval: ±5.353The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlExpert prompts Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 5,340 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 5,311 anonymous pairwise votes.	Winner1,490.882 95% confidence interval: ±8.777The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,466.306 95% confidence interval: ±8.867The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlFrench Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 1,472 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 1,966 anonymous pairwise votes.	Winner1,494.334 95% confidence interval: ±17.019The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,459.263 95% confidence interval: ±15.286The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlGerman Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 726 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 906 anonymous pairwise votes.	Winner1,485.2 95% confidence interval: ±22.822The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,433.595 95% confidence interval: ±20.539The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlHard prompts Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 31,069 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 34,700 anonymous pairwise votes.	Winner1,494.691 95% confidence interval: ±4.663The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,458.083 95% confidence interval: ±4.832The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlHard prompts, English Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 13,168 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 16,039 anonymous pairwise votes.	Winner1,490.87 95% confidence interval: ±6.176The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,464.61 95% confidence interval: ±6.133The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlInstruction following Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 16,713 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 18,051 anonymous pairwise votes.	Winner1,464.775 95% confidence interval: ±5.696The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,433.413 95% confidence interval: ±5.881The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlJapanese Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 639 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 635 anonymous pairwise votes.	Winner1,474.61 95% confidence interval: ±25.099The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,424.242 95% confidence interval: ±25.166The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlKorean Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 929 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 951 anonymous pairwise votes.	Winner1,435.275 95% confidence interval: ±20.203The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,378.153 95% confidence interval: ±20.872The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlLegal and government Test details No documented difference in the test setupNo clear leadVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 3,917 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 4,107 anonymous pairwise votes.	No clear lead1,476.168 95% confidence interval: ±10.17The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,457.491 95% confidence interval: ±9.927The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlLife, physical and social science Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 7,683 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 8,509 anonymous pairwise votes.	Winner1,488.692 95% confidence interval: ±7.391The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,459.724 95% confidence interval: ±7.418The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlLonger queries Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 21,881 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 23,501 anonymous pairwise votes.	Winner1,480.524 95% confidence interval: ±5.441The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,448.636 95% confidence interval: ±5.577The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlMath Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 2,261 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 2,619 anonymous pairwise votes.	Winner1,495.757 95% confidence interval: ±12.768The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,441.412 95% confidence interval: ±11.893The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlMathematical professions Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 2,730 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 2,828 anonymous pairwise votes.	Winner1,482.844 95% confidence interval: ±11.978The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,445.003 95% confidence interval: ±11.761The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlMedicine and healthcare Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 3,429 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 3,774 anonymous pairwise votes.	Winner1,484.201 95% confidence interval: ±10.913The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,461.307 95% confidence interval: ±10.453The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlMulti-turn Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 7,746 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 9,233 anonymous pairwise votes.	Winner1,481.739 95% confidence interval: ±7.534The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,446.332 95% confidence interval: ±7.34The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlNon-English Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 26,754 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 28,837 anonymous pairwise votes.	Winner1,468.08 95% confidence interval: ±4.816The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,423.125 95% confidence interval: ±5.023The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlPolish Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 765 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 1,037 anonymous pairwise votes.	Winner1,510.58 95% confidence interval: ±22.117The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,434.977 95% confidence interval: ±18.457The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlRussian Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 5,188 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 5,790 anonymous pairwise votes.	Winner1,488.319 95% confidence interval: ±8.821The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,437.327 95% confidence interval: ±8.593The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlSoftware and IT services Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 18,620 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 21,329 anonymous pairwise votes.	Winner1,501.513 95% confidence interval: ±5.465The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,472.722 95% confidence interval: ±5.539The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlSpanish Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 1,388 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 1,497 anonymous pairwise votes.	Winner1,470.651 95% confidence interval: ±16.935The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,429.802 95% confidence interval: ±16.45The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
LMArena Text, Style ControlWriting, literature and language Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGemini 3.5 Flash: Snapshot: gemini-3.5-flash-high, Reasoning: high, Harness: Arena text, style control DeepSeek-V4-Flash: Snapshot: deepseek-v4-flash-high-preview, Reasoning: high, Harness: Arena text, style controlGemini 3.5 Flash: Style-controlled Arena rating from 11,905 anonymous pairwise votes.DeepSeek-V4-Flash: Style-controlled Arena rating from 12,763 anonymous pairwise votes.	Winner1,472.184 95% confidence interval: ±6.479The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,422.277 95% confidence interval: ±6.549The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/25/2026
Key to the marks:better valuebetter, but not conclusivetie