AIEvals

Models

GPT-6.1 Sol

OpenAI’s GPT-6.1 Sol model for coding and agentic work, with published results from both Artificial Analysis and Vals AI.

OpenAIReleased 29 September 2026Context 1.05M tokensOutput 128k tokensPrivate weights$2.00 in / $10.00 out per Mtok

Family GPT-637 benchmarks with a published result1 independent first place among the 46 models this site listsRead 17 October 2026

Where it stands

Rank compares tracked models with a score. Published shows the source’s rank and field size, when available.

BenchmarkScoreRankPublishedLeaderBehind
Vals IndexVals AI61.15%8of 338of 41Gemini 4 Argon7.75 pts
Artificial Analysis Intelligence IndexArtificial Analysis526of 42no list readClaude Opus 5.56 pts
Terminal-Bench 4.0Artificial Analysis56.06%5of 40no list readClaude Sonnet 5.57.58 pts
Vibe Code Bench v1.1Vals AI88.93%7of 387of 106Claude Sonnet 5.53.46 pts
Terminal-Bench Science 0.1Artificial Analysis58.10%3of 24no list readGPT-6 Astra5.23 pts
MLCR-AAArtificial Analysis33.89%13of 37no list readClaude Sonnet 5.541.11 pts
AutomationBench-AAArtificial Analysis64.87%8of 40no list readGemini 4 Argon12.64 pts
SciCodeArtificial Analysis54.17%27of 40no list readClaude Opus 5.512.73 pts
GDP.pdfArtificial Analysis31.00%1of 40no list readGPT-6.1 Sol—
CritPtArtificial Analysis31.71%2of 42no list readGPT-5.6 Sol0.58 pts
AA-OmniscienceArtificial Analysis41.536of 42no list readClaude Opus 5.54.89
AA-LCR v1.1Artificial Analysis83.00%12of 42no list readKimi K35.67 pts
ProofBench v1.1Vals AI99.00%4of 345of 46Claude Sonnet 5.51.00 pts
Harvey's Legal Agent BenchmarkVals AI5.42%23of 3729of 73Muse Spark 1.220.00 pts
MysteryMechanismVals AI46.40%5of 215of 21GPT-6 Astra6.75 pts
BioMysteryBenchVals AI79.63%2of 192of 21Claude Sonnet 5.51.48 pts
Code MigrationVals AI65.12%5of 375of 71Claude Sonnet 5.54.71 pts
CyberBench v1.1Vals AI39.29%30of 3141of 43GPT-6 Sol38.69 pts
EMBVals AI70.77%12of 3712of 69Claude Fable 5.15.90 pts
Finance Agent (v2)Vals AI52.03%26of 3731of 73Gemini 4 Argon13.37 pts
IOIVals AI96.89%3of 313of 38Gemini 4 Argon3.11 pts
Legal Research BenchVals AI38.46%26of 3728of 72Muse Spark 1.3 Max16.83 pts
MedCodeVals AI48.84%15of 3526of 104Claude Opus 514.73 pts
MedScribeVals AI86.45%18of 3621of 106Claude Opus 5.54.98 pts
Public Benefits Bench v1.1Vals AI59.34%25of 2931of 45Claude Opus 517.59 pts
SAGEVals AI46.54%20of 3235of 90Claude Opus 4.79.56 pts
SRE BenchVals AI50.76%2of 152of 16GPT-6 Astra6.11 pts
Tax Agent BenchVals AI62.31%23of 3727of 64Claude Fable 5.115.33 pts
GDPval-AA v2.1Artificial Analysis1,575.0921of 42no list readClaude Opus 5.5271.08
AA-Briefcase v1.1Artificial Analysis1,564.2111of 41no list readClaude Opus 5.5257.64
Humanity's Last ExamArtificial Analysis52.92%8of 42no list readClaude Opus 5.58.43 pts

Nearby on Vals Index

The two models above and below on the Vals Index (Vals AI), on every column where this model has a published number.

ModelVals Index%Artificial Analysis Intelligence IndexindexTerminal-Bench 4.0%Vibe Code Bench v1.1%Terminal-Bench Science 0.1%
GPT-6 AstraOpenAI63.13%5359.09%89.59%63.33%
Claude Fable 5Anthropic61.39%5042.42%90.35%–
GPT-6.1 SolOpenAI61.15%5256.06%88.93%58.10%
Muse Spark 1.3 MaxMeta58.16%4833.33%85.86%10.95%
GPT-5.6 SolOpenAI58.01%4739.90%80.50%22.38%
Bold marks a figure better than this model’s on the same column, whichever model it belongs to. Ranks cover the models shown from the stated source.

Cost and speed, as published

Additional measurements from Artificial Analysis’ own row. Read 30 September 2026.

As publishedCost per taskOutput speed
GPT-6.1 Sol (max)$0.7269.3 tok/s

First answer token: 291.08 s; 500-token response: 298.30 s. Sol at max reasoning effort, measured with 10,000 input tokens. First-answer latency includes reasoning; Artificial Analysis reports its median over the past 72 hours. Output speed counts tokens generated after output begins. The response time is for 500 output tokens. These timings are separate from the Intelligence Index cost per task. Provider measurements

Results by benchmark

Independent and vendor results retain their own harnesses and effort settings.

BenchmarkIndependent runAs reported by a publisher
OSWorld 2.1 (offline strict pass)Computer useno published result
39.8%OSWorld 2.1 (offline strict pass)Anthropic · 7 October 2026Rank 3 of 6 in this publisher table
MLCR-AAMedical reasoning33.89%Artificial Analysisnot published
AutomationBench-AAWorkflow automation64.87%Artificial Analysisnot published
SciCodeScientific coding54.17%Artificial Analysisnot published
GDP.pdfDocument reasoning31.00%Artificial Analysisnot published
CritPtScientific reasoning31.71%Artificial Analysisnot published
AA-OmniscienceFactual knowledge41.53Artificial Analysisnot published
AA-LCR v1.1Long-context reasoning83.00%Artificial Analysisnot published
Terminal-Bench 4.0Agentic coding56.06%Artificial Analysis
55.05%Vals AI
58.2%Terminal-Bench 4.0Anthropic · 7 October 2026Rank 3 of 6 in this publisher table
MMMU ProMultimodal reasoning
85.95%Artificial Analysis
not published
Vibe Code Bench v1.1Product coding88.93%Vals AInot published
Terminal-Bench Science 0.1Agentic science58.10%Artificial Analysis
57.0%Terminal-Bench-Science 0.1Anthropic · 7 October 2026Rank 3 of 5 in this publisher table
57.0%Terminal-Bench Science 0.1OpenAI · 29 September 2026Rank 3 of 4 in this publisher table
ProofBench v1.1Mathematics99.00%Vals AInot published
Harvey's Legal Agent BenchmarkProfessional knowledge5.42%Vals AInot published
MysteryMechanismScientific reasoning46.40%Vals AInot published
Vals IndexComposite61.15%Vals AInot published
BioMysteryBenchAgentic science79.63%Vals AInot published
Code MigrationAgentic coding65.12%Vals AInot published
CyberBench v1.1Security39.29%Vals AInot published
EMBFinancial modelling70.77%Vals AInot published
Finance Agent (v2)Financial analysis52.03%Vals AInot published
IOICompetitive programming96.89%Vals AInot published
Legal Research BenchProfessional knowledge38.46%Vals AInot published
MedCodeHealthcare administration48.84%Vals AInot published
MedScribeHealthcare administration86.45%Vals AInot published
Public Benefits Bench v1.1Public services59.34%Vals AInot published
SAGEEducation46.54%Vals AInot published
SRE BenchSecurity50.76%Vals AInot published
Tax Agent BenchProfessional knowledge62.31%Vals AInot published
OSWorld 2.0Computer useno published result
71.4%OSWorld 2.0 (offline set, partial reward)OpenAI · 29 September 2026Rank 2 of 3 in this publisher table
OSWorld 2.1Computer useno published result
76.6%OSWorld 2.1 (offline partial)Anthropic · 7 October 2026Rank 3 of 6 in this publisher table
AutomationBenchBusiness workflowsno published result
36.1%AutomationBench 1.0.6OpenAI · 29 September 2026Rank 3 of 5 in this publisher table
DeepSWE v1.1Software engineeringno published result
75.2%DeepSWE v1.1OpenAI · 29 September 2026Rank 1 of 3 in this publisher table
GDPval-AA v2.1Knowledge work1,575.09Artificial Analysisnot published
AA-Briefcase v1.1Knowledge work1,564.21Artificial Analysisnot published
Humanity's Last ExamMultidisciplinary reasoning52.92%Artificial Analysisnot published
Artificial Analysis Intelligence IndexComposite52Artificial Analysisnot published
Independent column: the suite named under each figure, one harness per suite. Publisher column: each lab’s own table, reproduced on the benchmark’s page. Publisher ranks compare reported scores in the same table row only; ties share a rank and unreported cells are excluded.

Source notes

From the linked sources. Read 30 September 2026.

Token limits and input. 1,050,000 total context tokens; at most 922,000 input tokens and 128,000 output tokens. Text and image input; text output. Audio and video are unsupported. Knowledge cutoff: 30 April 2026.

Reasoning effort. low, medium (default), high, xhigh and max; none and minimal are unsupported.

Tool calling. Use the Responses API for tools. Chat Completions supports requests without tools. Streaming and structured outputs are supported; fine-tuning is unavailable.

Standard prices / Mtok. $2 input, $0.10 cached input, $2.50 cache writes and $10 output.

Long prompts. With more than 272,000 input tokens, the full request costs $4 input, $0.20 cached input, $5 cache writes and $15 output per Mtok.

Processing modes. Fast costs twice Standard; Batch and Flex cost half. Regional processing adds 10%. US and EU data residency are supported; Fast is unavailable in the EU.

Sources

Read on 17 October 2026.