OpenAIReasoning modelCompare with another model

GPT-6.1 Sol

Reasoning effort

GPT-6.1 Sol is an OpenAI proprietary reasoning model, released 29 Sept 2026. Its best configuration (max reasoning effort) ranks #7 of 775 on the BenchLeader Index at 70.0 ±3.3, in the top ten. The ± is the point: 43 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in composite (88) and lowest in agents & tools (65). At $4.00 per million tokens blended it is among the most expensive ranked models. Output speed of 55 tokens per second puts it slower than most, with a first answer in 248.2 s. It has been measured at 6 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 19 Oct 2026.

Blended price: $4.00/M; $2.00 in · $10.00 out
Output speed: 55 tok/s; measured by Artificial Analysis
First answer: 248 s; first token 3.39 s
Context: 1.1M
Full answer: 257 s; median, reasoning included
Cost per run: $0.724; one full Intelligence Index run
Released: 29 Sept 2026

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Overall index70
Reasoning76
Coding72
Agents & tools65
Maths73
Knowledge78
Human preference68
Multimodal66
Long context67
Composite88

Versions

OpenAI has shipped 3 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

Model	Released	Index	Rank
GPT-6.1 Solmaxthis page	29 Sept 2026	70.0	#7
GPT-6 Solmax	22 Sept 2026	66.6	#30
GPT-5.6 Solmax	9 Jul 2026	68.0	#16

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

Effort	Index	Rank	Speed	First answer	Chat reply cost	Categories
low	65.0	#41	43 tok/s	2.86 s	$0.0038	Coding 59 · Composite 77 · Knowledge 79 · Long context 67 · Multimodal 66 · Reasoning 73
medium	67.2	#27	45 tok/s	5.32 s	$0.0038	Coding 62 · Composite 83 · Knowledge 80 · Long context 67 · Multimodal 67 · Reasoning 76
high	67.7	#20	49 tok/s	50 s	$0.0038	Coding 62 · Composite 86 · Knowledge 80 · Long context 66 · Multimodal 68 · Reasoning 79
xhigh	67.7	#21	51 tok/s	175 s	$0.0038	Coding 62 · Composite 87 · Knowledge 80 · Long context 65 · Multimodal 68 · Reasoning 80
maxbest	70.0	#7	55 tok/s	248 s	$0.0038	Agents & tools 65 · Coding 72 · Composite 88 · Human preference 68 · Knowledge 78 · Long context 67 · Maths 73 · Multimodal 66 · Reasoning 76
not stated	63.3	#67	33 tok/s	3.39 s	$0.0038	Agents & tools 65 · Composite 70 · Instruction following 83 · Knowledge 54 · Maths 78

What more thinking costs

Turning the effort up buys index points and multiplies the bill. Here max costs 5.5× low for +5.0 on the index — 2.0 points per doubling of spend. Compare that with other models.

How its index has moved

2 Oct 2026 to 19 Oct 2026

max
high
xhigh
medium

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	low	medium	high	xhigh	max	not stated	Source
GPQA Diamond	–	–	–	–	95.4%#4	–	Epoch AI Benchmarking Hub
Arena Hard Prompts	–	–	–	–	1506#21	–	Arena (LMArena)
LiveBench 2.0 Reasoningnot in indexeffort not stated	–	–	–	–	–	92.6%#3	LiveBench
LiveBench 1.0 Reasoningnot in index	–	–	–	91.6%#6	92.6%#2	–	LiveBench
Humanity's Last Exam (AA)not in index	47.4%#39	49.9%#26	51.4%#22	52.6%#21	52.9%#18	–	Artificial Analysis
ARC-AGI-1	93.5%#46	95.5%#31	98.5%#1	98.5%#1	96.5%#23	–	ARC Prize
ARC-AGI-2	76.7%#46	86.7%#26	91.7%#9	91.7%#9	94.2%#2	–	ARC Prize
ARC-AGI-3	82.8%#14	91.0%#13	95.0%#12	96.4%#10	96.2%#11	–	ARC Prize
CritPt	24.9%#44	27.7%#29	30.0%#16	31.7%#3	31.7%#3	–	Artificial Analysis
MysteryMechanismnot in index	–	–	–	–	46.4%#5	–	Vals AI
Chess Puzzlesnot in index	–	–	–	–	68.0%#3	–	Epoch AI Benchmarking Hub
Mystery Game Puzzlesnot in index	–	–	–	–	80.0%#2	–	Epoch AI Benchmarking Hub
LMCAnot in index	–	–	–	–	62.8%#27	–	Epoch AI Benchmarking Hub
DTBenchnot in index	–	–	–	–	96.5%#22	–	Epoch AI Benchmarking Hub

Coding

Benchmark	low	medium	high	xhigh	max	not stated	Source
SciCode	53.2%#79	53.2%#79	55.8%#43	55.7%#45	54.2%#64	–	SciCode
FrontierCode	–	50.2%#7	–	–	–	–	Cognition
Arena Coding	–	–	–	–	1540#10	–	Arena (LMArena)
Arena WebDev	–	–	–	–	1753#4	–	Arena (LMArena)
LiveBench 2.0 Codingnot in indexeffort not stated	–	–	–	–	–	96.0%#4	LiveBench
LiveBench 1.0 Codingnot in index	–	–	–	80.7%#17	80.4%#20	–	LiveBench
SciCode (AA)not in index	53.2%#78	53.2%#78	55.8%#44	55.7%#46	54.2%#65	–	Artificial Analysis
IOI	–	–	–	–	96.9%#3	–	Vals AI
Code Migrationnot in index	–	–	–	–	65.1%#5	–	Vals AI
ProgramBenchnot in index	–	–	–	–	3.0%#5	–	Vals AI
Vibe Code Bench v1.1not in index	–	–	–	–	88.9%#8	–	Vals AI
ALE-Benchnot in index	–	–	–	–	2743.8#2	–	Epoch AI Benchmarking Hub
Terminal-Bench 4.0 (AA)not in index	30.8%#44	48.0%#21	51.5%#18	54.0%#12	56.1%#10	–	Artificial Analysis
Startrise Frontendnot in indexeffort not stated	–	–	–	–	–	78.6%#5	Startrise
FrontierCode 1.1 (AA)not in index	–	–	–	–	49.1%#5	–	Artificial Analysis

Agents & tools

Benchmark	low	medium	high	xhigh	max	not stated	Source
Terminal-Bench	–	–	–	–	58.2%#17	–	Terminal-Bench
APEX-Agents	–	–	–	–	60.0%#12	–	Mercor
Arena Agent	–	–	–	–	13.7#4	–	Arena (LMArena)
LiveBench Long Agenticnot in indexeffort not stated	–	–	–	–	–	73.0%#5	LiveBench
LiveBench Browsernot in indexeffort not stated	–	–	–	–	–	88.0%#7	LiveBench
LiveBench Agentic Codingnot in index	–	–	–	56.8%#25	54.5%#32	–	LiveBench
HiL-Bencheffort not stated	–	–	–	–	–	62.3%#4	Scale AI SEAL
GDPval-AA v2.1not in index	40.7%#100	47.5%#62	50.6%#49	51.9%#45	54.7%#35	–	Artificial Analysis
Analyst Agent (AA)not in index	–	–	–	–	50.0%#10	–	Artificial Analysis
CyberBenchnot in index	–	–	–	–	39.3%#43	–	Vals AI
Finance Agent v2not in index	–	–	–	–	52.0%#36	–	Vals AI
Harvey's Legal Agent Benchmarknot in index	–	–	–	–	6.3%#18	–	Vals AI
Legal Research Benchnot in index	–	–	–	–	38.5%#33	–	Vals AI
SREBenchnot in index	–	–	–	–	50.8%#2	–	Vals AI
Tax Agent Benchnot in index	–	–	–	–	21.8%#42	–	Vals AI
Terminal-Bench 4.0 (Vals)not in index	–	–	–	–	55.0%#7	–	Vals AI
Terminal-Bench Sciencenot in index	–	–	–	–	52.9%#2	–	Vals AI
Arena Searchnot in index	–	–	–	–	1250#2	–	Arena (LMArena)
AutomationBenchnot in index	52.6%#70	62.6%#25	64.5%#20	66.6%#14	64.9%#18	–	Artificial Analysis
GDP.pdf (AA)not in index	27.0%#17	30.0%#11	32.0%#3	31.8%#4	31.0%#5	–	Artificial Analysis
Harvey LAB-AA v1.1not in index	–	–	–	–	6.9%#6	–	Harvey
AA-Briefcase v1.1not in index	–	–	1465#35	1504#30	1557#21	–	Artificial Analysis
Terminal-Bench-Science 0.1 (AA)not in index	–	–	–	–	58.1%#4	–	Artificial Analysis
FACTS Searchnot in indexeffort not stated	–	–	–	–	–	88.2%#3	Google

Maths

Benchmark	xhigh	max	not stated	Source
FrontierMath Tiers 1–3	–	93.7%#1	–	Epoch AI Benchmarking Hub
FrontierMath Tier 4	–	100.0%#1	–	Epoch AI Benchmarking Hub
OTIS Mock AIME	–	100.0%#1	–	Epoch AI Benchmarking Hub
ProofBencheffort not stated	–	–	99.0%#4	Vals AI
LiveBench Mathematicsnot in index	96.5%#7	96.8%#3	–	LiveBench
Arena Maths	–	1494#24	–	Arena (LMArena)

Knowledge

Benchmark	low	medium	high	xhigh	max	not stated	Source
SimpleQA Verified	–	–	–	–	73.9%#3	–	Epoch AI Benchmarking Hub
LiveBench 2.0 Data Analysisnot in indexeffort not stated	–	–	–	–	–	82.7%#3	LiveBench
LiveBench Researchnot in indexeffort not stated	–	–	–	–	–	83.5%#11	LiveBench
LiveBench 1.0 Data Analysisnot in index	–	–	–	82.2%#3	82.7%#2	–	LiveBench
AA-Omniscience	37.6#20	40.0#18	41.5#12	40.9#13	41.5#11	–	Artificial Analysis
PRBench Financeeffort not stated	–	–	–	–	–	46.2%#19	Scale AI SEAL
PRBench Legaleffort not stated	–	–	–	–	–	47.9%#18	Scale AI SEAL
BioMysteryBenchnot in index	–	–	–	–	79.6%#2	–	Vals AI
Excel Modeling Benchmarknot in index	–	–	–	–	70.8%#13	–	Vals AI
MedCodenot in index	–	–	–	–	48.8%#28	–	Vals AI
MedScribenot in index	–	–	–	–	86.5%#23	–	Vals AI
Public Benefits Benchnot in index	–	–	–	–	59.3%#34	–	Vals AI
EBR-benchnot in index	–	–	–	–	54.3%#5	–	Epoch AI Benchmarking Hub
GDP.pdfnot in index	–	–	–	–	35.0%#1	–	Epoch AI Benchmarking Hub
AA-Omniscience: accuracynot in index	58.9%#25	60.4%#19	60.8%#17	60.8%#16	62.1%#12	–	Artificial Analysis
AA-Omniscience: non-hallucinationnot in index	48.4%#126	48.4%#127	50.6%#113	49.1%#120	45.7%#141	–	Artificial Analysis

Instruction following

Benchmark	xhigh	max	not stated	Source
LiveBench Languagenot in index	88.7%#6	90.1%#2	–	LiveBench
LiveBench Instruction Followingnot in index	71.3%#25	74.2%#12	–	LiveBench
MultiChallengeeffort not stated	–	–	82.0%#1	Scale AI SEAL
Arena Instruction Followingnot in index	–	1487#16	–	Arena (LMArena)

Human preference

Benchmark	max	Source
Arena Text	1484#22	Arena (LMArena)
Arena Creative Writingnot in index	1458#30	Arena (LMArena)
Arena Multi-turnnot in index	1486#27	Arena (LMArena)
Arena Longer Queriesnot in index	1492#25	Arena (LMArena)

Multimodal

Benchmark	low	medium	high	xhigh	max	not stated	Source
Arena Vision	–	–	–	–	1288#29	–	Arena (LMArena)
LiveBench Designnot in indexeffort not stated	–	–	–	–	–	82.1%#10	LiveBench
MMMU-Pro	83.1%#27	83.9%#23	84.9%#13	85.1%#11	86.0%#6	–	Artificial Analysis
SAGEnot in index	–	–	–	–	46.5%#35	–	Vals AI
AA-3Dnot in index	–	–	–	–	1118#7	–	Artificial Analysis

Long context

Benchmark	low	medium	high	xhigh	max	Source	Trend
AA-LCR	84.0%#11	83.3%#20	82.3%#40	79.7%#87	83.0%#25	Artificial Analysis
MLCRnot in index	–	–	–	–	33.9%#18	Artificial Analysis

Composite

Benchmark	low	medium	high	xhigh	max	not stated	Source
Epoch Capabilities Indexnot in indexeffort not stated	–	–	–	–	–	168.4#3	Epoch AI Benchmarking Hub
LiveBench 2.0effort not stated	–	–	–	–	–	85.4%#4	LiveBench
LiveBench 1.0not in index	–	–	–	81.1%#9	81.6%#6	–	LiveBench
AA Intelligence Index v4.3.2	42.1#53	47.8#26	50.2#19	51.0#16	51.8#13	–	Artificial Analysis
Vals Indexnot in index	–	–	–	–	63.9#7	–	Vals AI

Where it wins

Benchmarks where this configuration ranks in the top five of every configuration measured.

Where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

Provider	Speed	First token	Input $/M	Output $/M	Context	Quantisation
OpenAI	148 tok/s	9.82 s	$12.00	$60.00	1.1M	–
OpenAI	53 tok/s	2.06 s	$4.00	$20.00	1.1M	–
OpenAI	50 tok/s	3.68 s	$1.00	$5.00	1.1M	–
Amazon Bedrock	42 tok/s	3.07 s	$2.20	$11.00	1.1M	–
OpenAI	34 tok/s	3.59 s	$2.00	$10.00	1.1M	–
Azure	30 tok/s	3.85 s	$2.20	$11.00	1.1M	–
Azure	26 tok/s	5.31 s	$2.00	$10.00	1.1M	–
Azure	25 tok/s	3.79 s	$2.20	$11.00	1.1M	–

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.100 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0038	$0.0032	4.2 min
Summarize a 30-page report	12,000 / 600	$0.030	$0.013	4.3 min
Code edit	6,000 / 1,500	$0.027	$0.018	4.6 min
Agentic coding session	60,000 / 4,000	$0.160	$0.074	5.3 min
Structured extraction	2,000 / 200	$0.0060	$0.0032	4.2 min