GPT-6.1 Sol

  • Closed weights
  • Reasoning
  • Input: file · image · text
  • Released 2026-09
RankedEnough results for an index.
IndexAll categories, out of 100.76.2±2.2Median 50.7
CoverageIndex weight that has results.100%
SpeedOutput tokens per second.46/sMedian 76/s
First tokenSeconds to the first output token.2.9 sMedian 1.1 s
Input / 1MUS dollars per 1M input tokens.$2.00batch $1.00Median $0.60
Output / 1MUS dollars per 1M output tokens.$10.00batch $5.00 Batch price per 1M output tokens.Median $2.50
ContextMaximum tokens in one request.1.05MMedian 262K
EloLMArena rating and rank.1446±6 #63Median 1425

Details

From OpenRouter
API IDThe model ID on OpenRouter.
openai/gpt-6.1-sol
Max outputMaximum output tokens per request.
128K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.10
Weights
—
RetiresDate OpenRouter removes the model.
—

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.14 counted74.011 / 221
CodingCode writing and repair.11 counted74.113 / 214
ReasoningLogic problems and puzzles.10 counted80.15 / 225
MultimodalTasks with images and text.3 counted71.19 / 88
KnowledgeFacts and expert knowledge.8 counted76.47 / 266
MultilingualTasks in many languages.1 counted72.8—
InstructionTasks with strict prompt rules.1 counted75.0—
MathMath problems.5 counted76.62 / 171

Results

53 counted
BenchmarkResultThe score from the source.PlacePlace among results on this benchmark.
OTIS Mock AIME 2024-2025Math100.0%1 / 203
EuroEval ItalianMultilingual78.5% ±0.31 / 191
EuroEval SwedishMultilingual77.5% ±0.42 / 213
EuroEval PortugueseMultilingual74.8% ±0.42 / 199
EuroEval DutchMultilingual74.0% ±0.43 / 192
EuroEval PolishMultilingual66.9% ±0.34 / 227
EuroEval GermanMultilingual67.1% ±0.54 / 208
Critical Physics TasksReasoningAbove fitFar above this model's index.31.7%4 / 207
EuroEval FrenchMultilingual76.1% ±0.44 / 195
ARC-AGI-2 (semi-private)Reasoning94.2%3 / 137
EuroEval SpanishMultilingual68.2% ±0.55 / 194
LiveBench LanguageKnowledge90.1%2 / 66
GPQA diamond · Epoch AIKnowledge95.4% ±1.48 / 220
Artificial Analysis MMMU-ProMultimodal86.0%4 / 109
LiveBench ReasoningReasoning92.6%3 / 80
LiveBench Data AnalysisReasoning82.7%3 / 80
LiveBench MathematicsMath96.8%3 / 66
Mystery Game PuzzlesReasoningAbove fitFar above this model's index.80.0% ±4.05 / 83
Chess PuzzlesReasoning68.0% ±4.76 / 93
Code MigrationCoding65.1% ±4.45 / 77
IOICoding96.9% ±3.13 / 46
SimpleQA VerifiedKnowledge73.9% ±1.47 / 94
LiveBench Coding 2.0Coding96.0%4 / 52
LiveBench Long AgenticAgentic73.0%4 / 51
Vibe Code Bench v1.1Coding88.9% ±1.912 / 130
ARC-AGI-1 (semi-private)Reasoning96.5%14 / 132
ProofBench v1.1Math99.0% ±1.06 / 55
Furniture AssemblyReasoning80.0% ±4.95 / 40
Terminal-Bench 4.0.0Agentic58.2% ±1.63 / 23
LiveBench Browser UseAgentic88.0%8 / 49
GDPval-AA normalizedAgentic54.8%22 / 132
Terminal-Bench 4.1.0Agentic60.9% ±1.59 / 52
Terminal-Bench 4.0Agentic55.1% ±1.812 / 68
LiveBench Instruction FollowingInstruction74.2%12 / 66
Agent Arena task outcomeAgentic11.611 / 59
LiveBench DesignAgentic82.1%10 / 52
ARC-AGI-3 (semi-private)Reasoning52.7%6 / 30
LiveBench ResearchAgentic83.5%11 / 52
Agent Arena steerabilityAgentic11.513 / 59
PostTrainBench v1.2Coding38.8%6 / 23
ExploitGymAgentic35.1%6 / 23
Bug Hunt BenchCoding44.3%8 / 30
LiveBench CodingCoding80.4%18 / 66
FrontierCode 2.0Coding39.6%9 / 32
MirrorCodeCoding61.1% ±11.312 / 40
EBR-benchReasoning54.3% ±5.810 / 33
HealthBench HardKnowledge36.2%4 / 13
Artificial Analysis SciCodeCoding54.2%35 / 113
DeepSWEAgentic71.9%20 / 64
HealthBench ProfessionalKnowledge64.2%7 / 21
LiveBench Agentic CodingAgentic54.5%31 / 66
Terminal-Bench-Science 0.1Agentic57.0%9 / 15
AutomationBenchAgentic36.1%23 / 33
SRE BenchOther50.8% ±3.12 / 37
ARC-AGI-1 (public eval)Reasoning98.5%7 / 116
ARC-AGI-2 (public eval)Reasoning97.1%8 / 127
SAGEEducation90.5% ±1.52 / 30
ProgramBenchCoding3.0% ±1.26 / 66
Terminal-Bench-ScienceScience52.9% ±6.06 / 48
FrontierMath-ErdosMath4.4% ±2.52 / 16
BioMysteryBenchScience79.6% ±0.75 / 34
EMBFinance70.8% ±2.714 / 75
MysteryMechanismScience46.4% ±3.47 / 34
MedScribeHealthcare86.5% ±1.923 / 111
MedCodeHealthcare48.8% ±2.128 / 108
SpireClimbOther23.8% ±3.73 / 11
Legal Research BenchLegal38.5% ±3.444 / 98
Finance Agent (v2)Finance52.0% ±0.337 / 79
Tax Agent BenchFinance21.8% ±3.054 / 89
Public Benefits Bench v1.1Public benefits59.3% ±1.335 / 50
HLAB-Vals-1.1Legal6.3% ±1.831 / 42
CyberBench v1.1Other39.3% ±2.747 / 49
HealthBench raw scoreKnowledge56.7%—
Harvey LAB-AA v1.1Agentic6.9%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseEpoch AICC BYVals AINo licence statedLiveBenchNo licence statedARC Prize FoundationTerms restrict reuseEuroEvalMITTerminal-BenchNo licence statedLMArenaCC BY 4.0models.devMITCheaper InferenceNo licence stated

History

13 releases
Index79.3 → 76.2
23 Oct 202621 Nov 2026
Output / 1M$10.00 → $10.00
23 Oct 202621 Nov 2026

Questions

How much does GPT-6.1 Sol cost?
$2.00 per 1M input tokens, $10.00 per 1M output tokens.
How good is GPT-6.1 Sol?
Its index is 76.2 out of 100. That is place 8 of 282 ranked models.
What is GPT-6.1 Sol best at?
Math: 76.6 out of 100, place 2 of 171.
How fast is GPT-6.1 Sol?
46 output tokens per second, 2.9 s to the first token.
What is the context window of GPT-6.1 Sol?
1.05M tokens.
Is GPT-6.1 Sol open source?
No. Its weights are not published.
When was GPT-6.1 Sol released?
September 2026.

Badge

GPT-6.1 Sol on Inference360: place 8 of 282, index 76.2

Compare with

Gemini 4 Hydrogen76.2GPT-6.2 Sol76.0GPT-6 Astra76.5Muse Wonder Full75.8Claude Sonnet 5.575.7

Lower price

GPT-6.1 Terra75.1 · $5.00/1MGemini 4 Hydrogen76.2 · $7.50/1M

More from OpenAI

All models
GPT-6.1 Sol ProUnrankedGPT-6 Luna ProUnrankedGPT-6 Sol ProUnrankedGPT-6 Sol71.2GPT-6 Luna64.7