What are GPT-6 Sol's benchmark scores?

As of 9 Nov 2026, GPT-6 Sol (max effort) scores 62.9 ±3.2 on the BenchLeader Index, top 50, strongest in maths (69) and weakest in agents & tools (60). That rests on independent benchmark results only, and every score on this page links to its publisher.

How does GPT-6 Sol rank?

GPT-6 Sol (max effort) is #44 of 599 ranked configurations on 9 Nov 2026 (top 50). 136 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Fable 5.2 (max effort) at 69.4.

Is GPT-6 Sol better than GPT-5.3-Codex?

As of 9 Nov 2026, GPT-5.3-Codex (xhigh) and GPT-6 Sol (max) are neck and neck on quality (62.9 vs 62.9 on the BenchLeader Index, inside the ±7.6 margin of error), and GPT-6 Sol (max) costs 1.2× less, which makes it the better buy for most workloads. The head-to-head page has the benchmark-by-benchmark detail.

When was GPT-6 Sol released?

OpenAI released GPT-6 Sol on 22 Sep 2026. BenchLeader has results for it at 8 reasoning-effort settings.

OpenAIReasoning modelCompare with another model

GPT-6 Sol

Reasoning effort

Quality#19 of 335

62.9Index

Top 10%

+12.9 on the median

Price#204 of 251

$4.00per 1M tokens

Dearest quarter

4.7× the median

Speed#51 of 217

121tokens/s

Top quarter

1.6× the median

1 cheaper ranked model scores higher: GPT-6.1 Terra (65.0, $2.00).

As of 9 Nov 2026, GPT-6 Sol by OpenAI scores 62.9 on the BenchLeader Index (top 50 of 599 ranked configurations), with its best results in maths (69) and weakest in agents & tools (60). API pricing: $2.00 per 1M input tokens and $10.00 per 1M output tokens ($4.00 blended at 3:1), list price on 9 Nov 2026. It streams about 120 tokens/s (Artificial Analysis). Released 22 Sep 2026.

Key facts (as of 9 Nov 2026)

GPT-6 Sol API pricing: $2.00 per 1M input tokens and $10.00 per 1M output tokens ($4.00 blended at 3:1) (list price, 9 Nov 2026).
GPT-6 Sol streams about 120 tokens/s (Artificial Analysis, 9 Nov 2026).
GPT-6 Sol was released on 22 Sep 2026 (OpenAI).

In detail. GPT-6 Sol is an OpenAI proprietary reasoning model, released 22 Sep 2026. Its best configuration (max reasoning effort) ranks #44 of 599 on the BenchLeader Index at 62.9 ±3.2, in the upper half. The ± is the point: 136 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 maths (69) and lowest in agents & tools (60). At $4.00 per million tokens blended it is among the most expensive ranked models. Output speed of 121 tokens per second puts it in the fastest quarter, with a first answer in 72.1 s. It has been measured at 8 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 8 Nov 2026.

Blended price: $4.00/M; $2.00 in · $10.00 out; via Artificial Analysis
Output speed: 121 tok/s; measured by Artificial Analysis
First answer: 72 s; first token 2.30 s
Context: 1.1M; via models.dev
Full answer: 130 s; median, reasoning included
Cost per task: $2.47; 7 min a task, AA Index
Tokens per task: 54k; 66% of it reasoning
Released: 22 Sep 2026

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

Overall index63
Reasoning66
Coding67
Agents & tools60
Maths69
Knowledge62
Human preference64
Multimodal61
Long context67

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 Solmax	29 Sep 2026	67.0	#8
GPT-6 Solmaxthis page	22 Sep 2026	62.9	#44
GPT-5.6 Solmax	9 Jul 2026	65.4	#17

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
no reasoning	49.1	#366	114 tok/s	1.41 s	$0.0038	Coding 50 · Long context 56 · Multimodal 53 · Reasoning 36
low	54.6	#221	103 tok/s	1.74 s	$0.0038	Coding 54 · Long context 64 · Multimodal 63 · Reasoning 47
medium	57.0	#163	76 tok/s	2.30 s	$0.0038	Coding 59 · Long context 66 · Multimodal 63 · Reasoning 54
high	58.2	#135	108 tok/s	8.73 s	$0.0038	Coding 61 · Long context 67 · Multimodal 65 · Reasoning 58
xhigh	58.6	#123	116 tok/s	24 s	$0.0038	Coding 61 · Long context 65 · Multimodal 65 · Reasoning 60
maxbest	62.9	#44	121 tok/s	72 s	$0.0038	Agents & tools 60 · Coding 67 · Human preference 64 · Knowledge 62 · Long context 67 · Maths 69 · Multimodal 61 · Reasoning 66
pro mode	–	–	76 tok/s	2.30 s	$0.0038
not stated	–	–	76 tok/s	2.30 s	$0.0038	Knowledge 37 · Maths 70

How its index has moved

23 Sep 2026 to 9 Nov 2026

max
xhigh
high
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	no reasoning	low	medium	high	xhigh	max	Source
GPQA Diamond	–	–	–	–	–	94.3%#9	Epoch AI Benchmarking Hub
Arena Hard Prompts	–	–	–	–	–	1483#60	Arena (LMArena)
LiveBench 1.0 Reasoningnot in index	–	–	–	–	–	88.7%#22	LiveBench
Humanity's Last Exam (AA)not in index	18.4%#247	34.9%#135	41.0%#91	44.1%#59	46.3%#50	47.9%#41	Artificial Analysis
HLE-Diamond (AA)	9.5%#116	–	–	–	–	45.2%#28	Artificial Analysis
ARC-AGI-1	29.3%#216	72.2%#147	83.7%#122	91.0%#75	92.7%#59	95.5%#38	ARC Prize
ARC-AGI-2	1.7%#218	31.5%#136	57.8%#102	68.9%#67	78.1%#48	89.6%#20	ARC Prize
ARC-AGI-3	1.2%#45	1.8%#43	4.9%#35	5.9%#33	9.5%#28	23.0%#25	ARC Prize
CritPtnot in index	4.0%#185	16.3%#96	24.6%#55	25.4%#51	28.0%#33	30.9%#15	Artificial Analysis
CritPt v2 hard subset (AA)	–	–	–	–	–	66.5%#17	Artificial Analysis
MysteryMechanismnot in index	–	–	–	–	–	30.2%#15	Vals AI
Chess Puzzlesnot in index	–	–	–	–	–	56.0%#15	Epoch AI Benchmarking Hub
Mystery Game Puzzlesnot in index	–	–	–	–	–	56.0%#10	Epoch AI Benchmarking Hub
LMCAnot in index	–	–	–	–	–	59.1%#36	Epoch AI Benchmarking Hub
DTBenchnot in index	–	–	–	–	–	97.3%#9	Epoch AI Benchmarking Hub

Coding

Benchmark	no reasoning	low	medium	high	xhigh	max	not stated	Source
SciCode	47.3%#133	50.2%#114	53.8%#74	54.9%#61	55.1%#53	57.6%#29	–	SciCode
FrontierCode	–	–	–	–	–	49.3%#9	–	Cognition
Arena Coding	–	–	–	–	–	1520#38	–	Arena (LMArena)
Arena WebDev	–	–	–	–	–	1674#14	–	Arena (LMArena)
LiveBench 1.0 Codingnot in index	–	–	–	–	–	81.8%#13	–	LiveBench
SciCode (AA)not in index	47.3%#147	50.2%#121	53.8%#81	54.9%#67	55.1%#60	57.6%#32	–	Artificial Analysis
IOI	–	–	–	–	–	82.6%#10	–	Vals AI
Code Migrationnot in index	–	–	–	–	–	57.2%#9	–	Vals AI
ProgramBenchnot in index	–	–	–	–	–	2.0%#11	–	Vals AI
Vibe Code Bench v1.1not in index	–	–	–	–	–	87.8%#12	–	Vals AI
ALE-Benchnot in index	–	–	–	–	–	2462#4	–	Epoch AI Benchmarking Hub
Terminal-Bench 4.0 (AA)not in index	13.1%#99	9.1%#115	18.7%#87	26.3%#70	30.3%#62	43.9%#36	–	Artificial Analysis
GameBench 2 Overallnot in indexacross settings	–	–	–	–	–	–	72.9#8	DuelLab GameBench 2
GameBench 2 by reasoning settingnot in index	64.8#41	–	74.5#25	–	79.3#16	–	–	DuelLab GameBench 2
Design Arena Websitenot in indexeffort not stated	–	–	–	–	–	–	1285#31	Design Arena
FrontierCode 2.0 (AA)	–	–	–	–	–	35.6%#25	–	Artificial Analysis

Agents & tools

Benchmark	no reasoning	low	medium	high	xhigh	max	not stated	Source
Terminal-Bench	–	–	–	–	–	50.3%#33	–	Terminal-Bench
APEX-Agents	–	–	–	–	–	54.3%#26	–	Mercor
Arena Agent	–	–	–	–	–	7#12	–	Arena (LMArena)
LiveBench Agentic Codingnot in index	–	–	–	–	–	52.9%#37	–	LiveBench
Terminal-Bench 2.1 (Vals)	–	–	–	–	–	83.2%#6	–	Vals AI
ITBench SRE (AA)not in index	–	–	–	–	–	49.4%#7	–	Artificial Analysis
CyberBenchnot in index	–	–	–	–	–	78.0%#2	–	Vals AI
Finance Agent v2not in index	–	–	–	–	–	49.0%#49	–	Vals AI
Harvey's Legal Agent Benchmark v1not in index	–	–	–	–	–	1.7%#50	–	Vals AI
Legal Research Benchnot in index	–	–	–	–	–	28.9%#58	–	Vals AI
SREBenchnot in index	–	–	–	–	–	30.5%#6	–	Vals AI
Tax Agent Benchnot in index	–	–	–	–	–	15.0%#62	–	Vals AI
Terminal-Bench 4.0 (Vals)not in index	–	–	–	–	–	44.4%#14	–	Vals AI
Terminal-Bench Sciencenot in index	–	–	–	–	–	30.0%#10	–	Vals AI
Arena Searchnot in index	–	–	–	–	–	1229#4	–	Arena (LMArena)
Vending-Bench 2not in indexeffort not stated	–	–	–	–	–	–	14427.9#3	Andon Labs
AutomationBenchnot in index	34.2%#137	53.9%#84	58.0%#61	60.1%#49	61.7%#43	61.6%#44	–	Artificial Analysis
GDP.pdf (AA)not in index	17.0%#107	24.2%#47	23.8%#53	24.4%#46	24.6%#43	25.2%#40	–	Artificial Analysis
Harvey LAB-AA v1.1not in index	–	–	–	–	–	3.6%#16	–	Harvey
AA-Briefcase v1.1not in index	–	–	–	–	–	1476#49	–	Artificial Analysis
Design Arena Agenticnot in indexeffort not stated	–	–	–	–	–	–	1212#30	Design Arena
Terminal-Bench-Science 0.1 (AA)not in index	5.7%#73	3.3%#87	8.6%#66	17.1%#49	17.6%#47	30.0%#34	–	Artificial Analysis
CWE-Bench (AA)	–	–	–	–	–	64.2%#4	–	Artificial Analysis
DeepSec-Bench (AA)	–	–	–	–	–	46.3%#1	–	Artificial Analysis
CyberGym end-to-end (AA)	–	–	–	–	–	0.0%#19	–	Artificial Analysis

Maths

Benchmark	max	not stated	Source
FrontierMath Tiers 1–3	89.8%#6	–	Epoch AI Benchmarking Hub
FrontierMath Tier 4	90.0%#9	–	Epoch AI Benchmarking Hub
OTIS Mock AIME	100.0%#1	–	Epoch AI Benchmarking Hub
ProofBencheffort not stated	–	83.0%#13	Vals AI
LiveBench Mathematicsnot in index	96.4%#9	–	LiveBench
Arena Maths	1440#95	–	Arena (LMArena)

Knowledge

Benchmark	max	not stated	Source
SimpleQA Verified	60.7%#16	–	Epoch AI Benchmarking Hub
LiveBench 1.0 Data Analysisnot in index	81.2%#6	–	LiveBench
AA-Omniscience v1.1	21.4#50	–	Artificial Analysis
PRBench Legaleffort not stated	–	38.6%#36	Scale AI SEAL
BioMysteryBenchnot in index	74.8%#9	–	Vals AI
Excel Modeling Benchmarknot in index	71.5%#12	–	Vals AI
MedCodenot in index	47.1%#36	–	Vals AI
MedScribenot in index	82.0%#52	–	Vals AI
Public Benefits Bench v1.1not in index	56.6%#39	–	Vals AI
EBR-benchnot in index	53.3%#7	–	Epoch AI Benchmarking Hub
GDP.pdfnot in index	26.4%#15	–	Epoch AI Benchmarking Hub
AA-Omniscience v1.1: accuracynot in index	51.1%#29	–	Artificial Analysis
AA-Omniscience v1.1: non-hallucinationnot in index	39.1%#86	–	Artificial Analysis
HealthBench Professional (AA)not in index	62.3%#9	–	OpenAI

Instruction following

Benchmark	max	Source
LiveBench Languagenot in index	85.3%#15	LiveBench
LiveBench Instruction Followingnot in index	68.6%#35	LiveBench
Arena Instruction Followingnot in index	1448#68	Arena (LMArena)

Human preference

Benchmark	max	Source
Arena Text	1457#66	Arena (LMArena)
Arena Creative Writingnot in index	1430#74	Arena (LMArena)
Arena Multi-turnnot in index	1466#69	Arena (LMArena)
Arena Longer Queriesnot in index	1466#68	Arena (LMArena)

Multimodal

Benchmark	no reasoning	low	medium	high	xhigh	max	not stated	Source
Arena Vision	–	–	–	–	–	1232#76	–	Arena (LMArena)
MMMU-Pro	70.3%#156	80.2%#59	80.4%#57	82.0%#43	82.5%#38	83.0%#34	–	Artificial Analysis
Blueprint-Bench 2not in indexeffort not stated	–	–	–	–	–	–	36.9%#10	Epoch AI Benchmarking Hub
AA-3Dnot in index	–	–	–	–	–	1131#7	–	Artificial Analysis

Long context

Benchmark	no reasoning	low	medium	high	xhigh	max	Source	Trend
AA-LCR v1.1	64.0%#277	79.3%#110	82.3%#47	83.7%#17	81.3%#64	83.7%#17	Artificial Analysis
MLCRnot in index	22.8%#36	25.6%#30	28.9%#20	29.4%#19	32.2%#12	33.9%#9	Artificial Analysis

Composite

Benchmark	no reasoning	low	medium	high	xhigh	max	not stated	Source
Epoch Capabilities Indexnot in indexeffort not stated	–	–	–	–	–	–	164.1#10	Epoch AI Benchmarking Hub
LiveBench 1.0not in index	–	–	–	–	–	79.3%#14	–	LiveBench
AA Intelligence Index v5.0not in index	–	–	–	–	–	44.2#31	–	Artificial Analysis
AA Intelligence Index v4.3.2not in index	28.5#139	34.1#103	39.8#69	42.4#54	44.2#44	47.6#29	–	Artificial Analysis
AA Progress Index v1not in index	298.4#124	308.4#101	318.1#72	324.1#45	327.3#38	334#28	–	Artificial Analysis
Vals RSI Indexnot in index	–	–	–	–	–	28.1%#5	–	Vals AI
Vals Science Indexnot in index	–	–	–	–	–	45.0%#9	–	Vals AI
AA Cyber Indexnot in index	–	–	–	–	–	36.8#12	–	Artificial Analysis

Safety & honesty

Benchmark	not stated	Source	Trend
FORTRESSnot in index	13.5%#65	Scale AI SEAL

GameBench 2 Overall: 72.9. GameBench 2, best reasoning setting (XHigh): 79.3. Reference only, not part of the BenchLeader Index.

DuelLab GameBench 2, release gb2-20261103-eaea31df94e6 (https://benchmarks.duellab.org/releases/gb2-20261103-eaea31df94e6/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

Where it wins

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

API pricing: 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
Amazon Bedrock	85 tok/s	1.61 s	$2.00	$10.00	1.1M	–
OpenAI	83 tok/s	1.36 s	$1.00	$5.00	1.1M	–
Azure	80 tok/s	2.95 s	$2.20	$11.00	1.1M	–
OpenAI	73 tok/s	1.82 s	$4.00	$20.00	1.1M	–
OpenAI	68 tok/s	2.05 s	$2.00	$10.00	1.1M	–
Azure	45 tok/s	2.46 s	$2.00	$10.00	1.1M	–
Azure	41 tok/s	2.61 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 $0.200 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0038	$0.0033	1.2 min
Summarize a 30-page report	12,000 / 600	$0.030	$0.014	1.3 min
Code edit	6,000 / 1,500	$0.027	$0.019	1.4 min
Agentic coding session	60,000 / 4,000	$0.160	$0.079	1.8 min
Structured extraction	2,000 / 200	$0.0060	$0.0033	1.2 min

New this month

Compare GPT-6 Sol with…

Embed or cite

Show where GPT-6 Sol ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page. To see ranks like this on OpenRouter and Hugging Face model pages, get the Chrome extension.

[![GPT-6 Sol on BenchLeader: #19 overall](https://www.benchleader.com/badge/gpt-6-sol.svg)](https://www.benchleader.com/models/gpt-6-sol)

Cite this

The ranking changes daily, so a reference carries the data date. APA style:

BenchLeader. (2026, November 9). GPT-6 Sol on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 9, 2026, from https://www.benchleader.com/models/gpt-6-sol

BibTeX:

@misc{benchleader_gpt_6_sol_20261109,
  author       = {{BenchLeader}},
  title        = {{GPT-6 Sol on the BenchLeader Index (method v1.4)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/gpt-6-sol}},
  note         = {Data as of 2026-11-09}
}

Every badge style, the embeddable top-10 widget and how to cite the Index

Questions people ask

What are GPT-6 Sol's benchmark scores?: As of 9 Nov 2026, GPT-6 Sol (max effort) scores 62.9 ±3.2 on the BenchLeader Index, top 50, strongest in maths (69) and weakest in agents & tools (60). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does GPT-6 Sol rank?: GPT-6 Sol (max effort) is #44 of 599 ranked configurations on 9 Nov 2026 (top 50). 136 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Fable 5.2 (max effort) at 69.4.
How much does GPT-6 Sol cost per 1M tokens?: As of 9 Nov 2026, GPT-6 Sol costs $2.00 per 1M input tokens and $10.00 per 1M output tokens, $4.00 blended (three input tokens per output token).
Is GPT-6 Sol better than GPT-5.3-Codex?: As of 9 Nov 2026, GPT-5.3-Codex (xhigh) and GPT-6 Sol (max) are neck and neck on quality (62.9 vs 62.9 on the BenchLeader Index, inside the ±7.6 margin of error), and GPT-6 Sol (max) costs 1.2× less, which makes it the better buy for most workloads. The head-to-head page has the benchmark-by-benchmark detail.
When was GPT-6 Sol released?: OpenAI released GPT-6 Sol on 22 Sep 2026. BenchLeader has results for it at 8 reasoning-effort settings.

Data as of 9 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “GPT-6 Sol: benchmarks, pricing, speed and rank”, data as of 9 Nov 2026, https://www.benchleader.com/models/gpt-6-sol.