BenchLeader
OpenAIReasoning modelNewReleased 17 Nov 2026Compare with another model

GPT-6.2 Sol

Launch tracker · day 3

GPT-6.2 Sol at launch: the first independent results

As of 19 Nov 2026, GPT-6.2 Sol (OpenAI, released 17 Nov 2026) has independent results from Artificial Analysis (AA Intelligence Index v5.0.1 57.3), LMArena (Arena WebDev 1757), Vals AI (Vals Index 66.1), LiveBench (LiveBench 2.0 80.3%), ARC Prize (ARC-AGI-2 85.8%), Scale AI (MultiChallenge 72.0%), Kagi LLM Benchmark (Kagi LLM Benchmark 67.0%) and Terminal-Bench (Terminal-Bench 61.5%). It ranks #13 of 343 models on the BenchLeader Index, at 64.6. It is listed at $2.00 in / $10.00 out per 1M tokens, with a 1.1M-token context window. Not yet measured by Epoch AI and SWE-bench.

Released
17 Nov 2026 (day 3)
Provider
OpenAI
Price
$2.00 in / $10.00 out per 1M tokens
Context window
1.1M tokens
Open weights
No
BenchLeader Index
64.6 (#13 of 343 models)
Independent boardResults so far
Artificial AnalysisAA Intelligence Index v5.0.1 57.3 · AA-Omniscience v1.1 40.4 · HLE-Diamond (AA) 64.0% · more below
LMArenaArena WebDev 1757 · Arena Agent 11.3 · Arena Agent Task Completion 10.9
Vals AIVals Index 66.1 · BioMysteryBench 77.8% · Excel Modeling Benchmark v1.1 74.3% · more below
LiveBenchLiveBench 2.0 80.3% · LiveBench 2.0 Reasoning 90.6% · LiveBench 2.0 Coding 83.7% · more below
ARC PrizeARC-AGI-2 85.8% · ARC-AGI-1 97.5% · ARC-AGI-3 98.2%
Scale AIMultiChallenge 72.0% · PRBench Legal 43.3% · DrugDiscoveryBench 62.2% · more below
Kagi LLM BenchmarkKagi LLM Benchmark 67.0%
Terminal-BenchTerminal-Bench 61.5%

Boards usually post results within days of a launch; this page is rebuilt every three hours.

Scores from independent publishers only; benchmark figures a provider publishes about its own model are not shown.

Reasoning effort
Quality#11 of 351
64.6Index
Top 10%
+14.9 on the median
Price#210 of 262
$4.00per 1M tokens
Dearest quarter
4.5× the median
Speed#123 of 215
74tokens/s
Below the median
0.9× the median
Value#201 of 262
16.1points per $
Bottom quarter
0.3× the median

Better value: GPT-6.1 Terra scores 63.2 for $2.00/M, half the price.

No cheaper ranked model scores higher: it is on the price frontier.

As of 19 Nov 2026, GPT-6.2 Sol by OpenAI scores 64.6 on the BenchLeader Index (top 50 of 631 ranked configurations), with its best results in coding (75) and weakest in agents & tools (65). API pricing: $2.00 per 1M input tokens and $10.00 per 1M output tokens ($4.00 blended at 3:1), list price on 19 Nov 2026. It streams about 74 tokens/s (Artificial Analysis). Released 17 Nov 2026.

Key facts (as of 19 Nov 2026)

  • GPT-6.2 Sol API pricing: $2.00 per 1M input tokens and $10.00 per 1M output tokens ($4.00 blended at 3:1) (list price, 19 Nov 2026).
  • GPT-6.2 Sol streams about 74 tokens/s (Artificial Analysis, 19 Nov 2026).
  • GPT-6.2 Sol was released on 17 Nov 2026 (OpenAI).

In detail. GPT-6.2 Sol is an OpenAI proprietary reasoning model, released 17 Nov 2026. Its best configuration (max reasoning effort) ranks #27 of 631 on the BenchLeader Index at 64.6 ±3.0, in the upper half. The ± is the point: 92 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 coding (75) and lowest in agents & tools (65). At $4.00 per million tokens blended it is among the most expensive ranked models. Output speed of 74 tokens per second puts it slower than most, with a first answer in 78.8 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 18 Nov 2026.

Blended price
$4.00/M
$2.00 in · $10.00 out
via Artificial Analysis
Output speed
74 tok/s
measured by Artificial Analysis
First answer
79 s
first token 2.28 s
Context
1.1M
via models.dev
Accepts
Text · Images · PDFs
via models.dev, OpenRouter
Cost per task
$1.01
9 min a task, AA Index
Tokens per task
40k
56% of it reasoning
Released
17 Nov 2026
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index65
  2. Reasoning70
  3. Coding75
  4. Agents & tools65
  5. Knowledge65
  6. Composite65

Embed this rank

More badge styles
[![GPT-6.2 Sol on BenchLeader: #13 overall](https://www.benchleader.com/badge/gpt-6-2-sol.svg?style=compact)](https://www.benchleader.com/models/gpt-6-2-sol)

Paste it at the top of README.md, beside your other badges. The badge redraws with every data update and links back to this page.

Get an email when OpenAI ships a new model

And once when it gets ranked, plus the weekly review on Mondays.

Choose providers and emails

Versions

OpenAI has shipped 4 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.

ModelReleasedIndexRank
NewGPT-6.2 Solmaxthis page17 Nov 202664.6#27
GPT-6.1 Solmax29 Sep 202665.9#13
GPT-6 Solmax22 Sep 202662.4#56
GPT-5.6 Solmax9 Jul 202665.1#20

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.

EffortIndexRankSpeedFirst answerChat reply costCategories
low55.3#22263 tok/s2.04 s$0.0038Agents & tools 52 · Coding 60 · Knowledge 58 · Reasoning 58
medium61.6#6867 tok/s3.28 s$0.0038Agents & tools 63 · Coding 70 · Knowledge 62 · Reasoning 67
high62.3#5969 tok/s12 s$0.0038Agents & tools 64 · Coding 69 · Knowledge 64 · Reasoning 69
xhigh63.3#4573 tok/s32 s$0.0038Agents & tools 65 · Coding 72 · Knowledge 65 · Reasoning 70
maxbest64.6#2774 tok/s79 s$0.0038Agents & tools 65 · Coding 75 · Composite 65 · Knowledge 65 · Reasoning 70
not stated––51 tok/s2.28 s$0.0038Instruction following 66 · Knowledge 45 · Reasoning 61

What more thinking costs

Turning the effort up buys index points and multiplies the bill. Here max costs 5.5× low for +9.3 on the index — 3.8 points per doubling of spend. Compare that with other models.

566064low: $0.184 per task, index 55.355.3low $0.184medium: $0.406 per task, index 61.661.6medium $0.406high: $0.564 per task, index 62.362.3high $0.564xhigh: $0.648 per task, index 63.363.3xhigh $0.648max: $1.01 per task, index 64.664.6max $1.01Index

Each point is one reasoning setting: its cost per Artificial Analysis index task (log scale) and its BenchLeader Index.

How its index has moved

17 Nov 2026 to 19 Nov 2026
576268
  • 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

Benchmarklowmediumhighxhighmaxnot statedSourceTrend
LiveBench 2.0 Reasoningnot in index––––90.6%#12–LiveBench
HLE-Diamond (AA)34.5%#5850.7%#3758.3%#2661.8%#1864.0%#13–Artificial Analysis
Kagi LLM Benchmarkeffort not stated–––––67.0%#35Kagi LLM Benchmark
ARC-AGI-184.0%#13294.5%#5497.5%#1997.0%#3597.5%#19–ARC Prize
ARC-AGI-257.9%#11382.9%#5485.8%#4085.8%#4085.8%#40–ARC Prize
ARC-AGI-354.4%#3384.8%#3090.1%#2997.4%#1598.2%#9–ARC Prize
CritPt v2 hard subset (AA)58.0%#4764.0%#2461.5%#3162.5%#2662.5%#26–Artificial Analysis
MysteryMechanismnot in index––––36.9%#11–Vals AI
HLE-Diamond (Scale)not in indexeffort not stated–––––56.5%#4Scale AI SEAL

Coding

Benchmarklowmediumhighxhighmaxnot statedSourceTrend
Arena WebDev––––1757#8–Arena (LMArena)
LiveBench 2.0 Codingnot in index––––83.7%#11–LiveBench
Vibe Code Bench v1.1not in index––––87.6%#17–Vals AI
Vibe Code Bench 1-100 v1.1not in index––––40.9%#4–Vals AI
SWE Atlas v1.1not in indexeffort not stated–––––60.5%#16Scale AI SEAL
SWE Atlas v1.1: Codebase QnAnot in indexeffort not stated–––––61.3%#20Scale AI SEAL
SWE Atlas v1.1: Test Writingnot in indexeffort not stated–––––64.4%#10Scale AI SEAL
SWE Atlas v1.1: Refactoringnot in indexeffort not stated–––––55.7%#10Scale AI SEAL
Terminal-Bench 4.0 (AA)not in index37.9%#6356.1%#2558.6%#1956.1%#2560.1%#11–Artificial Analysis
FrontierCode 2.0 (AA)31.4%#4841.4%#1740.2%#2242.8%#1344.4%#10–Artificial Analysis

Agents & tools

BenchmarklowmediumhighxhighmaxSourceTrend
Terminal-Bench––––61.5%#20Terminal-Bench
Arena Agent––––11.3#11Arena (LMArena)
LiveBench Long Agenticnot in index––––63.6%#9LiveBench
LiveBench Browsernot in index––––88.2%#9LiveBench
Analyst Agent (AA)not in index38.8%#3550.0%#1952.5%#1453.8%#1053.8%#10Artificial Analysis
Finance Agent v2.1not in index––––58.9%#32Vals AI
Harvey's Legal Agent Benchmark v1.1not in index––––12.1%#22Vals AI
Legal Research Benchnot in index––––37.0%#53Vals AI
Tax Agent Benchnot in index––––24.8%#49Vals AI
Terminal-Bench 4.0 (Vals)not in index––––60.6%#8Vals AI
Terminal-Bench Sciencenot in index––––58.6%#4Vals AI
Arena Agent Task Completionnot in index––––10.9#13Arena (LMArena)
AutomationBenchnot in index63.2%#5670.3%#2669.8%#2770.5%#2571.3%#24Artificial Analysis
GDP.pdf (AA)not in index28.6%#2529.2%#2128.8%#2328.4%#2930.0%#16Artificial Analysis
Harvey LAB-AA v1.1not in index3.1%#285.0%#205.8%#186.1%#1711.1%#6Harvey
EnterpriseOps-Gymnot in index19.9%#6121.4%#6023.4%#5726.3%#5128.9%#47Artificial Analysis
AA-Briefcase v1.1not in index1230#1261495#631541#451580#321612#25Artificial Analysis
Terminal-Bench-Science 0.1 (AA)not in index37.1%#4360.5%#1867.6%#866.7%#1167.6%#8Artificial Analysis
CWE-Bench (AA)60.0%#1962.5%#1361.7%#1464.2%#961.7%#14Artificial Analysis
DeepSec-Bench (AA)27.9%#1445.5%#849.3%#448.9%#551.6%#3Artificial Analysis
CyberGym end-to-end (AA)not in index0.0%#220.0%#220.0%#220.0%#220.0%#22Artificial Analysis

Maths

BenchmarkmaxSourceTrend
Autoform-Hardnot in index85.7%#2Vals AI

Knowledge

Benchmarklowmediumhighxhighmaxnot statedSourceTrend
LiveBench 2.0 Data Analysisnot in index––––81.1%#12–LiveBench
LiveBench Researchnot in index––––90.1%#7–LiveBench
AA-Omniscience v1.124.8#6334.2#4738.3#4238.6#4140.4#33–Artificial Analysis
PRBench Legaleffort not stated–––––43.3%#33Scale AI SEAL
BioMysteryBenchnot in index––––77.8%#9–Vals AI
Excel Modeling Benchmark v1.1not in index––––74.3%#11–Vals AI
DrugDiscoveryBenchnot in index–––62.2%#7––Scale AI SEAL
AA-Omniscience v1.1: accuracynot in index54.6%#4159.2%#3461.4%#2260.7%#2761.6%#20–Artificial Analysis
AA-Omniscience v1.1: non-hallucinationnot in index34.2%#12438.9%#11340.0%#10843.8%#9744.6%#95–Artificial Analysis

Instruction following

Benchmarknot statedSourceTrend
MultiChallenge72.0%#10Scale AI SEAL

Multimodal

BenchmarkxhighmaxSourceTrend
LiveBench Designnot in index–80.0%#8LiveBench
Humanity's Sixth Sensenot in index56.8%#6–Scale AI SEAL
VIALS-AAnot in index–62.7%#4Handshake AI
AA-3Dnot in index–1259#3Artificial Analysis

Long context

BenchmarklowmediumhighxhighmaxSourceTrend
MLCRnot in index32.2%#2131.7%#2433.9%#1535.0%#1337.2%#9Artificial Analysis

Composite

BenchmarklowmediumhighxhighmaxSourceTrend
LiveBench 2.0––––80.3%#9LiveBench
AA Intelligence Index v5.0.1not in index40.5#5652.7#2754.9#1955.7#1757.3#13Artificial Analysis
Vals Indexnot in index––––66.1#10Vals AI
Vals Finance Indexnot in index––––56.7%#27Vals AI
Vals Science Indexnot in index––––57.8%#8Vals AI
Vals Legal Indexnot in index––––66.7%#34Vals AI
AA Cyber Indexnot in index29.3#2836.0#2437#2037.7#1837.8#17Artificial Analysis

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.

ProviderSpeedFirst tokenInput $/MOutput $/MContextQuantisation
Azure97 tok/s5.73 s$2.40$12.001.1M–
Amazon Bedrock87 tok/s1.01 s$2.00$10.001.1M–
Amazon Bedrock70 tok/s1.12 s$2.20$11.001.1M–
OpenAI69 tok/s4.05 s$4.00$20.001.1M–
OpenAI68 tok/s2.77 s$6.00$30.001.1M–
Azure65 tok/s2.51 s$2.00$10.001.1M–
Azure48 tok/s4.23 s$2.20$11.001.1M–
OpenAI44 tok/s2.70 s$2.00$10.001.1M–

Usage

How much GPT-6.2 Sol is used in practice, as the gateways that publish it report. Usage is not part of the Index. Most used models and labs.

Share of coding-agent turns, 30 days
0.01%
98% in Codex

Source: T3 Code via Slopalytics.

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.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0038$0.00321.4 min
Summarize a 30-page report12,000 / 600$0.030$0.0131.4 min
Code edit6,000 / 1,500$0.027$0.0181.7 min
Agentic coding session60,000 / 4,000$0.160$0.0742.2 min
Structured extraction2,000 / 200$0.0060$0.00321.4 min

More from OpenAI

New this month

Compare GPT-6.2 Sol with…

See also

Embed or cite

Show where GPT-6.2 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.

[![GPT-6.2 Sol on BenchLeader: #13 overall](https://www.benchleader.com/badge/gpt-6-2-sol.svg)](https://www.benchleader.com/models/gpt-6-2-sol)

Cite this

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

BenchLeader. (2026, November 19). GPT-6.2 Sol on the BenchLeader Index (method v1.5) [Data set]. Retrieved November 19, 2026, from https://www.benchleader.com/models/gpt-6-2-sol

BibTeX:

@misc{benchleader_gpt_6_2_sol_20261119,
  author       = {{BenchLeader}},
  title        = {{GPT-6.2 Sol on the BenchLeader Index (method v1.5)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/gpt-6-2-sol}},
  note         = {Data as of 2026-11-19}
}

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

Questions people ask

What are GPT-6.2 Sol's benchmark scores?
As of 19 Nov 2026, GPT-6.2 Sol (max effort) scores 64.6 ±3.0 on the BenchLeader Index, top 50, strongest in coding (75) and weakest in agents & tools (65). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does GPT-6.2 Sol rank?
GPT-6.2 Sol (max effort) is #27 of 631 ranked configurations on 19 Nov 2026 (top 50). 92 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 71.5.
How much does GPT-6.2 Sol cost per 1M tokens?
As of 19 Nov 2026, GPT-6.2 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-6.2 Sol better than Claude Opus 5?
As of 19 Nov 2026, Claude Opus 5 (max) and GPT-6.2 Sol (max) are neck and neck on quality (64.6 vs 64.6 on the BenchLeader Index, inside the ±3.6 margin of error), and GPT-6.2 Sol (max) costs 2.5× less, which makes it the better buy for most workloads. The head-to-head page has the benchmark-by-benchmark detail.
When was GPT-6.2 Sol released?
OpenAI released GPT-6.2 Sol on 17 Nov 2026. BenchLeader has results for it at 6 reasoning-effort settings.

Data as of 19 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “GPT-6.2 Sol: benchmarks, pricing, speed and rank”, data as of 19 Nov 2026, https://www.benchleader.com/models/gpt-6-2-sol.