OpenAILanguage modelActive

GPT-6 Sol

GPT-6 Sol is among the strongest models: Rank 13 of 92 in the overall index. Through the API it is fairly expensive.

IntelligenceRank 13 of 92

110.5Overall indexhigher is better

SpeedRank 33 of 100

111Tokens per secondhigher is better

PriceRank 71 of 95

Input $2 · Output $10

$4per 1M tokenslower is better

ContextRank 2 of 102

1.05Mtokens at once≈ 1,580 A4 pages

PopularityRank 3 of 785

73out of 100+1 vs. last month

based on search, usage and attention

Summary

It is fairly expensive: $4 per 1M tokens (input and output blended); a typical model costs $1.55. At 111 tokens per second it writes faster than a typical model (87).

GPT-6 Sol brings Astra’s training methods to a lower-cost tier. At launch, OpenAI halved standard input and output token prices compared with GPT-5.6 Sol’s promotional rates. Its internal factuality evaluation also found roughly half as many mistakes as with its predecessor. That test deliberately uses difficult conversations where users had previously flagged errors.

GPT-6.1 Sol later improves coding and computer-use results in the same line. GPT-6 Luna launched alongside GPT-6 Sol as the smaller, cheaper tier.

GPT-6 Sol benchmarks

Overall index (Gradually Meta-Score v4)

20 of 92 models

4 models in this view are measured at a lower reasoning effort than max in the overall index. Dashed column: lower reasoning effort than the top.

Claude Opus 5.5max126.1

Gemini 4 Carbonhigh125.8

Claude Fable 5.1max123.8

Claude Sonnet 5.5max122.8

Gemini 4 Argonhigh122.3

GPT-6 Astramax121.3

GPT-6.1 Solmax119.2

Muse Wonder Fullmax118.3

Claude Opus 5max117.2

Muse Wondermax113.4

Kimi K3.1 Promax112.9

Claude Fable 5max112.5

GPT-6 Solmax110.5

GPT-5.6 Solmax108.7

Muse Spark 1.3max107.4

Gemini 3.7 Flashhigh107.1

DeepSeek V4.1 Flash 1017max106.6

Gemini 3.8 Flashhigh106.5

GPT-6.1 Terramax105.3

Claude Haiku 5.5max103.4

GPT-6 Sol

DeepSeek V4.1 Flash 1017

Gemini 3.8 Flash

GPT-6.1 Terra

Claude Haiku 5.5

Speed

20 of 100 models

Mercury 2.5576tokens/s

Gemini 3.1 Flash-Lite336tokens/s

Gemini 3.5 Flash-Lite299tokens/s

Nemotron 3.5 Lightning245tokens/s

Claude Haiku 5.5236tokens/s

o3-mini230tokens/s

Ling 3.0 Flash199tokens/s

DeepSeek-V3 0324195tokens/s

GPT-5.4 mini191tokens/s

Gemini 3 Flash Preview191tokens/s

Gemini 3.5 Flash184tokens/s

Gemini 3.6 Flash182tokens/s

Command A Plus177tokens/s

DeepSeek-V4.1-Flash176tokens/s

DeepSeek V4.1 Flash 1017172tokens/s

Qwen 3.7 Max165tokens/s

GPT-5.4 nano153tokens/s

GPT OSS 20B152tokens/s

GPT OSS 120B151tokens/s

GPT-6 Sol111tokens/s

Mercury 2.5

Gemini 3.1 Flash-Lite

Gemini 3.5 Flash-Lite

Nemotron 3.5 Lightning

Gemini 3 Flash Preview

DeepSeek V4.1 Flash 1017

GPT-6 Sol

Output tokens per second, measured by Artificial Analysis

Intelligence vs. price

most attractive area: stronger and cheaper than half of the models shownbest intelligence for the price

GPT-6 Sol by task area

Points, 100 = reference group average · higher is better

MathematicsRank 4 of 47117.5

CodingRank 11 of 109114.9

ReasoningRank 10 of 72114.1

GeneralRank 9 of 65113.0

KnowledgeRank 8 of 32109.8

AgentsRank 16 of 98107.2

MultimodalRank 11 of 47106.9

HealthRank 13 of 72104.8

Legaltoo little data104.8

FinanceRank 30 of 104101.3

Light bar: too few results for a rank yet.

Text Arena

How people rate the answers in blind comparisons, Elo points with style control · higher is better

20 of 40 models

Gemini 4 Argon1,520

Claude Opus 5.51,509

Muse Wonder1,507

Claude Fable 5.11,504

Claude Fable 51,504

Gemini 3.8 Flash1,496

Muse Spark 1.31,495

Claude Opus 51,491

Gemini 3.7 Flash1,490

Kimi K31,488

Gemini 3.1 Pro Preview1,487

GPT-5.6 Sol1,485

Gemini 3.6 Flash1,483

GPT-6.1 Sol1,481

GLM-5.31,481

Gemini 3.5 Flash1,478

Claude Sonnet 5.51,477

GLM-5.3-Flash1,477

DeepSeek-V4.1-Flash1,476

GPT-6 Sol1,458

Gemini 3.1 Pro Preview

GPT-6 Sol

as of October 30, 2026 · Source: LMArena

WebDev Arena

How people rate generated websites and web apps in blind comparisons, Elo points · higher is better

20 of 43 models

Claude Opus 5.51,810

GPT-6 Astra1,773

Claude Sonnet 5.51,770

GPT-6.1 Sol1,751

Claude Fable 5.11,746

Muse Wonder Full1,725

Claude Opus 51,691

Gemini 4 Argon1,679

GPT-6 Sol1,678

Qwen 3.8 Max 09021,676

DeepSeek V4.1 Flash 10171,665

Kimi K31,656

Muse Spark 1.31,656

Muse Wonder1,652

Grok 4.71,637

Claude Fable 51,624

GLM-5.31,623

DeepSeek-V4.1-Flash1,619

Grok 4.61,616

GPT-5.6 Sol1,616

GPT-6 Sol

DeepSeek V4.1 Flash 1017

as of October 30, 2026 · Source: LMArena

Terminal-Bench 4.0

Share of tasks the model solves as an agent in the terminal, in percent · higher is better

20 of 38 models

Claude Opus 5.565.2 %

Claude Sonnet 5.564.1 %

Kimi K3.1 Pro63.6 %

Gemini 4 Carbon63.1 %

GPT-6 Astra59.6 %

Claude Fable 5.158.1 %

Muse Wonder Full58.1 %

Gemini 4 Argon57.6 %

GPT-6.1 Sol55.1 %

Claude Opus 553.5 %

GPT-6.1 Terra49.5 %

Muse Wonder49.5 %

GPT-6 Sol44.4 %

Claude Fable 541.4 %

GLM-5.338.9 %

GPT-5.6 Sol37.9 %

GLM-5.3-Flash37.4 %

Claude Haiku 5.536.4 %

Qwen 3.8 Max 090234.3 %

DeepSeek V4.1 Flash 101731.8 %

GPT-6 Sol

DeepSeek V4.1 Flash 1017

as of November 7, 2026 · Source: Vals AI

Vibe Code Bench

Share of apps the model builds correctly from a description, in percent · higher is better

20 of 51 models

Claude Sonnet 5.592.4 %

Gemini 4 Argon91.9 %

Claude Haiku 5.590.4 %

Claude Fable 590.4 %

Claude Opus 5.590.3 %

Claude Fable 5.190.3 %

GPT-6 Astra89.6 %

Gemini 4 Carbon89.1 %

GPT-6.1 Sol88.9 %

Claude Opus 588.4 %

GPT-6 Sol87.8 %

DeepSeek V4.1 Flash 101787.7 %

Grok 4.786.2 %

Muse Wonder Full86.0 %

Muse Spark 1.385.9 %

Kimi K385.0 %

Kimi K3.1 Pro84.9 %

DeepSeek-V4.1-Flash84.7 %

GPT-6.1 Terra84.0 %

Muse Wonder83.1 %

GPT-6 Sol

DeepSeek V4.1 Flash 1017

as of November 7, 2026 · Source: Vals AI

Show all 744 published benchmark results

Published benchmarks

Coding

LMArena Code Arena WebDev

HTML development

Note (GPT-6 Sol): The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.

Claude Fable 5.11,682.8 rating

GPT-6 Sol1,679.63 rating

Claude Fable 5.1

GPT-6 Sol

higher is betterAs of October 30, 2026Scale: 0 rating - 1,682.8 rating

Source: LMArena

gradually.ai

Methodology and context

Comparison cohort: lmarena-code-webdev:2026-10-28:webdev-html
Metric: WebDev Arena Bradley-Terry rating
Scale: No upper bound is documented. The chart spans the recorded values and zero.
Participants: 2
Methodology and context: Arena rating from 1,028 published pairwise votes.. Source model ID: lmarena/GPT-6 Sol (Max). Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Model snapshot: GPT-6 Sol (Max). Evaluator: LMArena. Dataset type: public-crowdsourced-preferences. Leaderboard date: 2026-10-28. Harness: Arena Code Arena WebDev. Reasoning effort: max. Source ID: lmarena. Evidence capture ID: lmarena-webdev-323062f70843f40e3f739865c3fe9d806126e412-4819e4823b65. Source payload hash: 4819e4823b65f309c40ca8ae73130851342e33621e89f30db1c284a7c0d89102. Parser version: lmarena-dataset-v1. Imported at: 2026-10-30

Source: LMArena

Code Migration

Sub-task	GPT-6 Sol (This model)	GPT-6 Astra	GPT-6.1 Sol	GPT-5.6 Sol	GPT-5.6 Terra	GPT-6 Luna	GPT-5.6 Luna
CLI	52.93%	67.46% (best value in this row)	63.5%	47.22%	40.41%	37.22%	36.07%
COBOL	70% (best value in this row)	68.57%	70% (best value in this row)	70% (best value in this row)	70% (best value in this row)	58.57%	70% (best value in this row)
Overall	57.2%	67.74% (best value in this row)	65.12%	52.92%	47.8%	42.55%	44.55%

higher is betterAs of November 7, 2026Source: Vals AIbars scaled within each row, not across rows

Methodology and context

Metric: accuracy

Code Migration

Code Quality

GPT-6.1 Sol4.5 rating

GPT-6 Astra4.47 rating

GPT-6 Sol4.38 rating

GPT-5.6 Terra3.95 rating

GPT-5.6 Sol3.85 rating

GPT-6 Luna3.75 rating

GPT-5.6 Luna3.2 rating

GPT-6 Sol

higher is betterAs of November 7, 2026Scale: 0 rating - 4.5 rating

Source: Vals AI

gradually.ai

Methodology and context

Comparison cohort: vals-code-migration:1:code_quality
Metric: mean reviewer code-quality rating
Scale: No upper bound is documented. The chart spans the recorded values and zero.
Participants: 7
Methodology and context: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-29. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 5136.155. Standard error: 0.046. Cost per task in USD: 15.676945. Score display: 4.375 / 5. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-11-07-23c368426e3e. Source payload hash: 23c368426e3e145dbba17219be86d157ba6ebf3b5c25c8f7ba2d5723986fb9b1. Parser version: vals-astro-v4. Imported at: 2026-11-07

Source: Vals AI

General tasks

BioMysteryBench

Sub-task	GPT-6 Sol (This model)	GPT-6.1 Sol	GPT-6 Astra	GPT-5.6 Sol	GPT-5.6 Luna	GPT-6 Luna
Hard	35.29%	39.22% (best value in this row)	37.26%	29.41%	27.45%	17.65%
Human-solvable	84.02%	89.04% (best value in this row)	89.04% (best value in this row)	80.82%	69.41%	71.69%
Overall	74.82%	79.63% (best value in this row)	79.26%	71.11%	61.48%	61.48%

higher is betterAs of November 7, 2026Source: Vals AIbars scaled within each row, not across rows

Methodology and context

Metric: accuracy

Cybersecurity

CyberBench v1.1

Sub-task	GPT-6 Sol (This model)	GPT-5.6 Sol	GPT-6 Luna	GPT-5.6 Luna	GPT-5.6 Terra	GPT-6 Astra	GPT-6.1 Sol
Overall	77.98% (best value in this row)	76.31%	76.25%	73.63%	72.08%	41.07%	39.29%
Patch	89.29% (best value in this row)	89.29% (best value in this row)	87.5%	83.93%	87.5%	82.14%	78.57%
PoC	66.67% (best value in this row)	63.33%	65%	63.33%	56.67%	0%	0%

higher is betterAs of November 7, 2026Source: Vals AIbars scaled within each row, not across rows

Methodology and context

Metric: accuracy

Recorded measurement details

LMArena Code Arena WebDev: HTML development (Show measurement, test conditions, and source)

Source value: 1679.6279514642752
Score: 1679.6279514642752
Metric: WebDev Arena Bradley-Terry rating
Unit: rating
Benchmark version: webdev
Category: coding
Task: webdev-html
Task label: HTML development
Direction: Higher is better
Comparison cohort: lmarena-code-webdev:2026-10-28:webdev-html
Confidence interval: 21.58371
Source type: official-leaderboard
Evaluator: LMArena
Status: active
Retrieved at: 2026-10-30
Methodology: Arena rating from 1,028 published pairwise votes.. Source model ID: lmarena/GPT-6 Sol (Max). Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Model snapshot: GPT-6 Sol (Max). Evaluator: LMArena. Dataset type: public-crowdsourced-preferences. Leaderboard date: 2026-10-28. Harness: Arena Code Arena WebDev. Reasoning effort: max. Source ID: lmarena. Evidence capture ID: lmarena-webdev-323062f70843f40e3f739865c3fe9d806126e412-4819e4823b65. Source payload hash: 4819e4823b65f309c40ca8ae73130851342e33621e89f30db1c284a7c0d89102. Parser version: lmarena-dataset-v1. Imported at: 2026-10-30
Comparison methodology: Arena rating from 1,028 published pairwise votes.. Harness: Arena Code Arena WebDev. Reasoning effort: max
Context: The Arena score measures human preference rather than a fixed capability test. The confidence interval, Arena category, and model configuration are part of the result.

Source LMArena

BioMysteryBench: Hard (Show measurement, test conditions, and source)

Source value: 35.294
Score: 35.294
Metric: accuracy
Unit: %
Benchmark version: 1
Category: general
Task: human_difficult
Task label: Hard
Direction: Higher is better
Comparison cohort: vals-biomysterybench:1:human_difficult
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-11-05. Harness: Terminus 2. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 502.413. Standard error: 0. Cost per task in USD: 1.113779. Source ID: vals-ai. Evidence capture ID: vals-biomysterybench-2026-11-07-c2992129e13e. Source payload hash: c2992129e13e1455796b51fe354f18fa049ad9dcfe02fd66196230a60aeea9df. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Harness: Terminus 2. Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

BioMysteryBench: Human-solvable (Show measurement, test conditions, and source)

Source value: 84.018
Score: 84.018
Metric: accuracy
Unit: %
Benchmark version: 1
Category: general
Task: human_solvable
Task label: Human-solvable
Direction: Higher is better
Comparison cohort: vals-biomysterybench:1:human_solvable
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-11-05. Harness: Terminus 2. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 283.024. Standard error: 2.994. Cost per task in USD: 0.557233. Source ID: vals-ai. Evidence capture ID: vals-biomysterybench-2026-11-07-c2992129e13e. Source payload hash: c2992129e13e1455796b51fe354f18fa049ad9dcfe02fd66196230a60aeea9df. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Harness: Terminus 2. Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

BioMysteryBench: Overall (Show measurement, test conditions, and source)

Source value: 74.815
Score: 74.815
Metric: accuracy
Unit: %
Benchmark version: 1
Category: general
Task: overall
Task label: Overall
Direction: Higher is better
Comparison cohort: vals-biomysterybench:1:overall
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-11-05. Harness: Terminus 2. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 324.465. Standard error: 2.429. Cost per task in USD: 0.662358. Source ID: vals-ai. Evidence capture ID: vals-biomysterybench-2026-11-07-c2992129e13e. Source payload hash: c2992129e13e1455796b51fe354f18fa049ad9dcfe02fd66196230a60aeea9df. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Harness: Terminus 2. Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

Code Migration: CLI (Show measurement, test conditions, and source)

Source value: 52.927
Score: 52.927
Metric: accuracy
Unit: %
Benchmark version: 1
Category: coding
Task: cli
Task label: CLI
Direction: Higher is better
Comparison cohort: vals-code-migration:1:cli
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-29. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 6436.976. Standard error: 2.372. Cost per task in USD: 20.385725. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-11-07-23c368426e3e. Source payload hash: 23c368426e3e145dbba17219be86d157ba6ebf3b5c25c8f7ba2d5723986fb9b1. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

Code Migration: COBOL (Show measurement, test conditions, and source)

Source value: 70
Score: 70
Metric: accuracy
Unit: %
Benchmark version: 1
Category: coding
Task: cobol
Task label: COBOL
Direction: Higher is better
Comparison cohort: vals-code-migration:1:cobol
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-29. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 1233.694. Standard error: 15.275. Cost per task in USD: 1.550605. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-11-07-23c368426e3e. Source payload hash: 23c368426e3e145dbba17219be86d157ba6ebf3b5c25c8f7ba2d5723986fb9b1. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

Code Migration: Overall (Show measurement, test conditions, and source)

Source value: 57.195
Score: 57.195
Metric: accuracy
Unit: %
Benchmark version: 1
Category: coding
Task: overall
Task label: Overall
Direction: Higher is better
Comparison cohort: vals-code-migration:1:overall
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-29. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 5136.155. Standard error: 4.213. Cost per task in USD: 15.676945. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-11-07-23c368426e3e. Source payload hash: 23c368426e3e145dbba17219be86d157ba6ebf3b5c25c8f7ba2d5723986fb9b1. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

Code Migration: Code Quality (Show measurement, test conditions, and source)

Source value: 4.375
Score: 4.375
Metric: mean reviewer code-quality rating
Unit: rating
Benchmark version: 1
Category: coding
Task: code_quality
Task label: Code Quality
Direction: Higher is better
Comparison cohort: vals-code-migration:1:code_quality
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-10-29. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 5136.155. Standard error: 0.046. Cost per task in USD: 15.676945. Score display: 4.375 / 5. Source ID: vals-ai. Evidence capture ID: vals-code-migration-2026-11-07-23c368426e3e. Source payload hash: 23c368426e3e145dbba17219be86d157ba6ebf3b5c25c8f7ba2d5723986fb9b1. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

CyberBench v1.1: Overall (Show measurement, test conditions, and source)

Source value: 77.976
Score: 77.976
Metric: accuracy
Unit: %
Benchmark version: 1.1
Category: cybersecurity
Task: overall
Task label: Overall
Direction: Higher is better
Comparison cohort: vals-cyberbench:1.1:overall
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-05. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 734.454. Standard error: 5.11. Cost per task in USD: 1.338572. Source ID: vals-ai. Evidence capture ID: vals-cyber-2026-11-07-cc5fe09bf1f8. Source payload hash: cc5fe09bf1f8c926613f8893f823a902fb33f219303590d6a2794e43708d67a5. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

CyberBench v1.1: Patch (Show measurement, test conditions, and source)

Source value: 89.286
Score: 89.286
Metric: accuracy
Unit: %
Benchmark version: 1.1
Category: cybersecurity
Task: patch
Task label: Patch
Direction: Higher is better
Comparison cohort: vals-cyberbench:1.1:patch
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-05. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 686.527. Standard error: 4.133. Cost per task in USD: 0.738255. Source ID: vals-ai. Evidence capture ID: vals-cyber-2026-11-07-cc5fe09bf1f8. Source payload hash: cc5fe09bf1f8c926613f8893f823a902fb33f219303590d6a2794e43708d67a5. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

CyberBench v1.1: PoC (Show measurement, test conditions, and source)

Source value: 66.667
Score: 66.667
Metric: accuracy
Unit: %
Benchmark version: 1.1
Category: cybersecurity
Task: poc
Task label: PoC
Direction: Higher is better
Comparison cohort: vals-cyberbench:1.1:poc
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-11-07
Methodology: Source model ID: openai/gpt-6-sol. Source model name: GPT-6 Sol. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-05. Reasoning effort: max. Maximum output tokens: 128000. Latency in seconds: 782.382. Standard error: 6.086. Cost per task in USD: 1.93889. Source ID: vals-ai. Evidence capture ID: vals-cyber-2026-11-07-cc5fe09bf1f8. Source payload hash: cc5fe09bf1f8c926613f8893f823a902fb33f219303590d6a2794e43708d67a5. Parser version: vals-astro-v4. Imported at: 2026-11-07
Comparison methodology: Reasoning effort: max. Maximum output tokens: 128000

Source Vals AI

Page 1 of 64

See for yourself: one prompt, one attempt

We show the latest attempt without fixes; it does not count toward the overall index.

Pelican on a bicycle

Prompt

Generate an SVG of a pelican riding a bicycle

Oct 1, 2026 · 80 s · $0.04

Test details

Run date: Oct 1, 2026
API provider: openrouter
Requested API model: openai/gpt-6-sol
Reasoning setting: high
Token limit including reasoning: 8,192 tokens
Output tokens: 4,389 tokens
Cost of the attempt: $0.0439
Response time: 80.5 s
API model reported in response: openai/gpt-6-sol
Pinned endpoint: OpenAI | openai/gpt-6-sol-20260922
Temperature: Provider default
Test protocol: community-visual-tests-v1
Technical checks: Passed, not a quality rating

Visually reviewed. This individual test does not establish overall model quality.

Task origin: Simon Willison

SVG short test

One SVG task with the same fixed settings for every model, without hidden thinking.

GPT-6 Sol

Unavailable in this round

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

The 100 most popular models of this category, 0 to 100, higher is more popular

20 of 100 models

GPT-6 Astra83

Claude Haiku 5.577

GPT-6 Sol73

GPT-573

DeepSeek V4.1 Flash 101769

Kimi K367

MiniCPM-V 4.7 35B-A3B63

GPT-6 Sol

DeepSeek V4.1 Flash 1017

MiniCPM-V 4.7 35B-A3B

As of November 2026 · Source: gradually.ai popularity index

Over time

Weekly values from October 18, 2026 to November 7, 2026

GPT-6 SolClaude Haiku 5.5 77DeepSeek V4.1 Flash 1017 69Claude Opus 5.5 68Qwen 3.8 27B 68

Method changes in the timeline, a jump at these points is no market event:

1November 1, 2026: Value only with search or usage data, 30-day attention

What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.

GPT-6 Sol compared

Compare with another model

First modelSecond model

Predecessor and successor

SuccessorGPT-6 Sol vs. GPT-6.1 SolSuccessor of GPT-6 Sol

The most popular comparisons

Show all 56 comparisons

Similar models

More models of the same kind as GPT-6 Sol, chosen by model line, developer, size, price and release date.

All specifications

Architecture

Knowledge cutoff: April 20, 2026; Source: Knowledge cutoff

Input and output

Context window: 1,050,000 tokens; Source: Context window
Maximum output: 128,000 tokens; Source: Maximum output
Input: Text, Image; Source: Input
Output: Text; Source: Output
Reasoning mode: none, low, medium, high, xhigh, max; Source: Reasoning mode
Tool calling: Yes; Source: Tool calling

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking: Unknown; no source
Watermarking technique: Not known; no source

Licence and access

Weights: Proprietary; Source: Weights
Access: API; Source: Access
Release: Sep 22, 2026; Source: Release

API

API model ID: gpt-6-sol; Source: API model ID

Further details

Notes: Released September 22, 2026 as the second GPT-6 tier below GPT-6 Astra, built by OpenAI for complex coding and agentic workflows. 1.05 million token context window, 128,000 maximum output tokens, and an April 20, 2026 knowledge cutoff. API price $2 / $10 per million input and output tokens, half of GPT-5.6 Sol's promotional price. Reasoning effort from none to max, with medium as the default. Official API ID: gpt-6-sol.; Source: Notes

Prices

API pricesPrice historySubscriptions

8 offers, prices in US dollars per 1M tokens, blended from 3 parts input and 1 part output.

Provider	Input	Output	Cache read	Cache write	Blended
OpenAIdirect	$2	$10	$0.20	$2.50	$4
OpenAIabove 272,000 context tokensdirect	$4	$15	$0.40	$5	$6.75
OpenAIFlex	$1	$5	n/a	n/a	$2
Amazon BedrockGlobal	$2	$10	n/a	n/a	$4
Azure	$2	$10	n/a	n/a	$4
Amazon BedrockUs-east-1	$2.20	$11	n/a	n/a	$4.40
AzureEU	$2.20	$11	n/a	n/a	$4.40
AzureUS	$2.20	$11	n/a	n/a	$4.40
OpenAIFast	$4	$20	n/a	n/a	$8

Cache prices come from the direct tariff; OpenRouter does not report them for the other providers. Direct price according to the developer, checked Sep 22, 2026. Provider prices via OpenRouter, as of Nov 7, 2026. Open the LLM price index

2 price changes since tracking began.

InputOutput

Daily observed price in the LLM price index in US dollars per 1M tokens. The index uses a market price from several sources, which can differ from the developer's direct price.

At OpenAI from $20 a month (ChatGPT Plus). Through other providers from $6 a month (ElevenLabs Starter, with limited credits). 57 plans in Germany.

Plan	Monthly list price	Note
ChatGPT PlusOpenAI	$20	developer's own plan
ChatGPT Business Standard-PlatzOpenAI	$25	developer's own plan, with limited credit, for teams, per seat
ChatGPT Pro $100OpenAI	$100	developer's own plan, with limited credit
ChatGPT Business Premium-PlatzOpenAI	$125	developer's own plan, with limited credit, for teams, per seat
ChatGPT EnterpriseOpenAI	n/a	developer's own plan, for enterprises, per seat
ChatGPT Pro $500OpenAI	$500	developer's own plan, with limited credit
ElevenLabs Free	free	with limited credit
Genspark Free	free	with limited credit
JetBrains AI Free	free	with limited credit
Lovable Free	free	with limited credit
Zed Student	free
ElevenLabs Starter	$6	with limited credit
Poe Basic	$4.99	with limited credit
Zed Pro	$10
JetBrains AI Pro	$10	with limited credit
GitHub Copilot Business	$19	with limited credit, for teams, per seat
Command Code Pro	$20	with limited credit
Replit Core	$20
ElevenLabs Creator	$22	with limited credit
JetBrains AI Pro	$20	with limited credit, for teams, per seat
Lovable Pro	$25	with limited credit
Genspark Plus	$24.99	with limited credit
Poe Plus	$19.99	with limited credit
Langdock Business Standard	€25	for teams, per seat
JetBrains AI Ultimate	$30	with limited credit
GitHub Copilot Enterprise	$39	with limited credit, for enterprises, per seat
GitHub Copilot Pro+	$39	with limited credit
Genspark for Business Team	$40	with limited credit, for teams, per seat
Lovable Business	$50	with limited credit, for teams
Mindverse Studio Team	€49	for teams, per seat
Poe Pro	$49.99	with limited credit
Higgsfield Plus	€59	with limited credit
Mindverse Studio Pro	€59	per seat
JetBrains AI Ultimate	$60	with limited credit, for teams, per seat
Higgsfield Team	€79	with limited credit, for teams, per seat
ElevenLabs Pro	$99	with limited credit
Command Code Max 10×	$100	with limited credit
GitHub Copilot Max	$100	with limited credit
Replit Pro	$100
Krea Max	$105	with limited credit
Langdock Business Max	€99	for teams, per seat
Poe Advanced	$99.99	with limited credit
Higgsfield Ultra	€129	with limited credit
Higgsfield Scale	€169	with limited credit, for teams, per seat
Command Code Max 20×	$200	with limited credit
Krea Business	$200	with limited credit, for teams
Genspark Pro	$249.99	with limited credit
ElevenLabs Scale	$299	with limited credit, for teams
Poe Max	$249.99	with limited credit
ElevenLabs Business	$990	with limited credit, for teams
ElevenLabs Enterprise	n/a	for enterprises
Genspark for Business Enterprise	n/a	with limited credit, for enterprises, per seat
Higgsfield Enterprise	n/a	for enterprises
JetBrains AI Enterprise	n/a	with limited credit, for enterprises, per seat
Krea Enterprise	n/a	for enterprises
Lovable Enterprise	n/a	for enterprises
Replit Enterprise	n/a	for enterprises

All plans in the AI subscription comparison

Frequently asked questions about GPT-6 Sol

How good is GPT-6 Sol in benchmarks?

Good: Rank 13 of 92 in the overall index.

Overall index (Gradually Meta-Score v4) from 40 benchmark results · data through Nov 7, 2026

IntelligenceOverall index in points, 100 = reference group average · higher is better

Muse Wonder113.4

Kimi K3.1 Pro112.9

Claude Fable 5112.5

GPT-6 SolRank 13110.5

GPT-5.6 Sol108.7

Muse Spark 1.3107.4

Gemini 3.7 Flash107.1

What is the Overall index (Gradually Meta-Score v4)?

A point value that combines many independent benchmarks into one number. 100 points is the average of a fixed reference group, higher is better. GPT-6 Sol scores 110.5 points, which is rank 13 of 92.

Own analysis by gradually.ai, method benchmark-meta-score-v4 · for GPT-6 Sol from 40 test series

Which data: Only current, reviewed results from independent evaluators such as Artificial Analysis, Vals AI, LMArena, LiveBench and Epoch AI. Vendor claims and historical or disputed values do not count.
One value per test: For each source and test series the latest retrieval with one fixed setup counts. When a source tests several reasoning levels, the highest documented one applies. A series needs at least 5 models.
Making tests comparable: Percentages are converted so that gaps near 0 and near 100 % count fairly, then every series is put on the same scale. A hard test therefore does not weigh more than an easy one.
Shared strength: A statistical model estimates, from all series at once, one shared capability per model and, per area, how far a model deviates from it. Missing tests are never counted as zero.
Weighting: Core areas (Coding, Reasoning, Mathematics, Knowledge, Agents, General and Multimodal) count 75 % together, the domain areas Finance, Legal and Health 25 %.
The point scale: The value is measured against a fixed reference group of 29 models, much like an IQ score: 100 points is its average and 15 points one standard deviation of the group. The scale is open at both ends and is not a percentage. A model 15 points ahead does better in roughly three of four benchmark comparisons. New models do not shift older values.
Place by score: The place follows the order of the scores among all models with a place, such as "rank 14 of 95". A second calculation checks whether a lead is reliable by recomputing the score 200 times with randomly redrawn test families. If a model is not almost always ahead there, the two models count as level.
Minimum evidence: A value needs at least 8 test series from 2 sources and 3 areas, a place only with enough precision. Missing core areas are flagged as "estimated".
What it is not: Not a test run by gradually.ai and no verdict on every single task. For a specific task the area values and the individual benchmarks matter more.

How much does GPT-6 Sol cost?

Through the API $2 per 1M input tokens and $10 output. That is fairly expensive.

Prices from Nov 7, 2026

Output price since tracking beganDollars per 1M tokens · lower is better

Sep 22, 2026Nov 7, 2026

First price on Sep 22, 2026: $10. Since Sep 24, 2026: $10.

What is the knowledge cutoff of GPT-6 Sol?

Apr 20, 2026, according to OpenAI. That is 5 months before release.

Stated by OpenAI

AprMayJunJulAugSepKnowledge cutoffApr 20, 2026ReleasedSep 22, 2026

How large is the context window of GPT-6 Sol?

1.05 M tokens. That is how much text GPT-6 Sol reads at once, about 1,580 A4 pages.

Page count as a rule of thumb from Artificial Analysis

Context windowTokens the model reads at once · higher is better

GPT-6 SolRank 2 of 1021.05M

GPT-5.6 Sol1.05M

Kimi K3.1 Pro1.05M

Muse Spark 1.31.05M

Muse Wonder1.05M

Gemini 3.7 Flash1.05M

Claude Fable 51M

How fast is GPT-6 Sol?

111 tokens per second. That is faster than a typical model at 87.

Measured by Artificial Analysis on Nov 6, 2026. AA 3.0.0, including time to first token, averaged across available workloads and five days.

SpeedTokens per second · higher is better

Muse Wonder124

Muse Spark 1.3120

GPT-6 SolRank 33 of 100111

GPT-5.6 Sol101

Claude Fable 586

Claude Opus 566

Kimi K3.1 Pro33

Can I run GPT-6 Sol locally?

No. OpenAI does not publish the weights.

Stated by OpenAI

Open weights: no
License: n/a
Parameters: n/a
active per token: n/a

Where can I use GPT-6 Sol?

At OpenAI with a subscription from $20 a month. Including other providers that makes 57 plans in Germany, plus the API.

Subscription comparison, prices in Germany incl. VAT

At OpenAI

6 plans, from $20 a month (ChatGPT Plus).

Through other providers

51 plans from 13 providers, from $6 a month (ElevenLabs Starter), with limited credits.

Through the API

$2 input, $10 output per 1M tokens, unchanged since Sep 24, 2026.

Sources and data date

Show all 103 sources

Benchmark sources at a glance

Which independent evaluators tested this model. This is coverage, not a quality ranking.

Evaluator	Test series	Composite indices	Recorded active measurements
Artificial Analysis	12	0	24
Vals AI	20	1	149
LMArena	4	0	42
LiveBench	0	0	38
Epoch AI	6	0	15
ARC Prize	0	0	24
Scale AI Evals	0	0	8

Without vendor claims and historical or disputed results; those stay labelled in the single values. 80 measurements are not yet assigned to a test series.

Type	Evidence and data date
Speed	OpenAI, Max effort. AA 3.0.0, including time to first token, averaged across available workloads and five days. Retrieved November 6, 2026. Source: Artificial Analysis · Speed (Artificial Analysis)Speed measurement methodology
Price	Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. Source: Pricing · Context window data (retrieved September 22, 2026) · GPT-6 Sol API model documentation · Editorial description reviewed October 4, 2026
Input	Documented modalities. Retrieved September 22, 2026. Source: Model documentation · Context window data (retrieved September 22, 2026) · GPT-6 Sol API model documentation · Editorial description reviewed October 4, 2026
Output	Documented modalities. Retrieved September 22, 2026. Source: Model documentation · Context window data (retrieved September 22, 2026) · GPT-6 Sol API model documentation · Editorial description reviewed October 4, 2026
Research status	Research date September 22, 2026. 69 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Not found in the inspected sources	parameters-billions Original wording OpenAI does not publish parameter counts.
Unresolved	fresh-agent-arena-leaderboard Note The current Agent Arena leaderboard was reviewed in the rendered web reader, but the raw endpoint returned HTTP 403. No dynamic values were added without a complete captured source body.
Not published	immutable-serving-revision Note The evaluator names the exact gpt-6-sol route and dated environment but does not publish the underlying immutable serving snapshot or exact hidden runtime version.
Not published	serving-revision-and-test-denominators Note ARC Prize pages do not state per-row N, CI, or immutable served revision. Parallel omits model snapshot and varies reasoning/token limits by provider guidance.
Not published	PRBench row subset and model-specific run settings Note The current leaderboard page does not identify whether this row uses the full or hard subset, or publish the model-specific generation settings or measurement date.
Additional source	Model metadata (retrieved September 22, 2026; October 19, 2026) · Introducing GPT-6 Sol and GPT-6 Luna · Editorial description reviewed October 4, 2026
Additional source	Artificial Analysis (retrieved October 19, 2026; November 7, 2026)1 2 3 4 5 6 7
Additional source	Artificial Analysis (retrieved October 19, 2026)1 2 3
Additional source	Artificial Analysis (retrieved November 7, 2026)1 2 3
Additional source	Artificial Analysis (retrieved October 19, 2026; November 2, 2026)
Additional source	Artificial Analysis (retrieved October 19, 2026; October 29, 2026; November 2, 2026; November 7, 2026)
Additional source	Artificial Analysis (retrieved October 19, 2026; November 2, 2026; November 7, 2026)1 2
Additional source	LMArena (retrieved October 30, 2026)1 2 3 4
Additional source	LMArena (retrieved October 28, 2026)1 2
Additional source	LMArena (retrieved October 16, 2026)1 2 3 4
Additional source	Vals AI (retrieved November 7, 2026)1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Additional source	Vals AI (retrieved October 16, 2026)1 2 3
Additional source	Vals AI (retrieved October 16, 2026; October 31, 2026; November 7, 2026)
Additional source	Vals AI (retrieved October 17, 2026; November 7, 2026)
Additional source	Vals AI (retrieved October 16, 2026; October 30, 2026; October 31, 2026)
Additional source	LiveBench (retrieved October 16, 2026)1 (unavailable on November 2, 2026) https://livebench.ai/table_2026_09_02.csv2 (unavailable on November 2, 2026) https://livebench.ai/breakdown_2026_09_02.json
Additional source	Arena AI (retrieved October 17, 2026)
Additional source	Joonlab GPT-6.1 Sol benchmark, pinned reasoning summary (retrieved October 17, 2026)
Additional source	ARC Prize Verified Leaderboard (retrieved October 17, 2026)
Additional source	Parallel Search Capability Leaderboard (retrieved October 17, 2026)
Additional source	Android Developers, Android Bench leaderboard (retrieved October 17, 2026)
Additional source	VetEval, GPT-6 Sol model page (retrieved October 17, 2026)
Additional source	Madrid-García and Merino-Barbancho, Jev in Medicine: A Benchmark Evaluation (retrieved October 17, 2026)
Additional source	BizBench research team, long-horizon business-agent benchmark (retrieved October 17, 2026)
Additional source	Joonlab GPT-6.1 Sol benchmark, pinned data commit (retrieved October 17, 2026)1 2 3
Additional source	Epoch AI, AI systems are improving rapidly at Slay the Spire and its sequel (retrieved October 17, 2026)
Additional source	VeriContest paper authors (retrieved October 18, 2026)
Additional source	Epoch AI benchmark run dataset snapshot (retrieved October 19, 2026; October 28, 2026)
Additional source	Roboflow Vision Evals (retrieved October 19, 2026; October 28, 2026)
Additional source	Agent Arena (retrieved October 20, 2026)
Additional source	VulcanBench (retrieved October 20, 2026)
Additional source	DuelLab GameBench 2 (retrieved October 20, 2026; October 28, 2026; November 2, 2026; November 5, 2026)
Additional source	TrueStandard (retrieved October 20, 2026)
Additional source	Towards AI ToneBench (retrieved October 20, 2026; October 28, 2026; October 30, 2026)
Additional source	SonarSource (retrieved October 20, 2026)
Additional source	OpenAI (retrieved October 23, 2026)
Additional source	OpenAI Deployment Safety Hub, GPT-6.1 Sol system-card addendum (retrieved October 23, 2026)
Additional source	OpenAI, GPT-6.1 Sol System Card Addendum (retrieved October 23, 2026)
Additional source	Scale AI (retrieved October 28, 2026)
Additional source	Google DeepMind on Kaggle (retrieved October 28, 2026)
Additional source	Google on Kaggle (retrieved October 28, 2026)1 2 3 4 5 6
Additional source	Terminal-Bench (retrieved October 28, 2026)
Additional source	Parallel Search Capability Leaderboard (retrieved October 28, 2026; November 5, 2026)
Additional source	Towards AI ToneBench (retrieved October 28, 2026; October 30, 2026)
Additional source	Scale AI Labs (retrieved October 28, 2026)
Additional source	Scale AI HSS (retrieved October 28, 2026)
Additional source	Scale AI DistressBench (retrieved October 28, 2026)
Additional source	Scale AI FORTRESS (retrieved October 28, 2026)
Additional source	Parallel Search Capability Leaderboard (retrieved November 5, 2026)
Additional source	OpenRouter: Popularity index (derived), CC BY 4.0
Additional source	Vercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional source	Ollama: cumulative downloads for the popularity index
Additional source	Replicate: cumulative runs for the popularity index
Additional source	Google Trending Now: limited feed selection for Germany and the US
Additional source	Creative Commons Attribution 4.0