Benchmark Almanac

Vals Science Index

GPT-6.2 Astra 68.4% leads since November 2026
GPT-6.2 Astra 68.4% leads since November 2026
In use

Agents are scored on the equal average of BioMysteryBench, MysteryMechanism and Terminal-Bench-Science.

Version 1.0 released October 16th, 2026. The board had 23 models on October 25th, 2026.

Made by
Vals AI
Released
October 2026
Kind
Composite indices
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
Kept private
Measures
Capability
Language
English
Interaction
Agentic
Environment
Terminal, Code sandbox
Items from
Another benchmark
Run by
A host
Strongest model tested
GPT-6.2 Astra #1 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 19th, 2026
28
Models scored on its own leaderboard
2
Launch posts list it, from 1 lab

GPT-6.2 Astra leads with 68.4%

Best score per model from the benchmark's own leaderboard, read on November 19th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

GPT-6.2 Astra leads with 68.4%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

  1. 1 GPT-6.2 Astra 68.4%
  2. 2 GPT-6 Astra 65.1%
  3. 3 Claude Fable 5.2 63.7%
  4. 4 Gemini 4 Neon 62.6%
  5. 5 GPT-6.1 Sol 59.6%
  6. 6 Claude Sonnet 5.5 59.1%
  7. 7 Claude Opus 5.5 58.6%
  8. 8 GPT-6.2 Sol 57.8%
  9. 9 Gemini 4 Argon 55.4%
  10. 10 Muse Wonder Full 48.5%
  11. 11 Claude Opus 5 47.9%
  12. 12 GPT-6 Sol 45%
  13. 13 Kimi K3.1 Pro 44.4%
  14. 14 GPT-5.6 Sol 41.5%
  15. 15 Muse Wonder 39.1%
  16. 16 Claude Haiku 5.5 38.5%
  17. 17 Grok 4.6 38.1%
  18. 18 Gemini 3.8 Flash 35.8%
  19. 19 Grok 4.7 34.8%
  20. 20 Inkling 2 Small 32.7%
  21. 21 Ember-1 31.8%
  22. 22 MiMo-V2.6 Flash 31.7%
  23. 23 DeepSeek V4.1 Flash 29.6%
  24. 24 Step 5 Preview 29.6%
  25. 25 GPT-6 Luna 28.4%
  26. 26 Gemini 3.1 Pro 27.4%
  27. 27 GPT-5.6 Luna 25.3%
  28. 28 Beam 1.0 22%

Source: Official Vals Science Index leaderboard

#ModelScore ± 95% CISettingDateSource
1 GPT-6.2 Astra 68.4% ±4.2 Nov 2026 Official leaderboard
2 GPT-6 Astra 65.1% ±4.4 Sep 2026 Official leaderboard
3 Claude Fable 5.2 63.7% ±4.5 Nov 2026 Official leaderboard
4 Gemini 4 Neon 62.6% ±4.6 Nov 2026 Official leaderboard
5 GPT-6.1 Sol 59.6% ±4.5 Sep 2026 Official leaderboard
6 Claude Sonnet 5.5 59.1% ±4.5 Sep 2026 Official leaderboard
7 Claude Opus 5.5 58.6% ±4.5 Sep 2026 Official leaderboard
8 GPT-6.2 Sol 57.8% ±4.4 Nov 2026 Official leaderboard
9 Gemini 4 Argon 55.4% ±4.5 Sep 2026 Official leaderboard
10 Muse Wonder Full 48.5% ±4.4 Oct 2026 Official leaderboard
11 Claude Opus 5 47.9% ±4.3 Jul 2026 Official leaderboard
12 GPT-6 Sol 45% ±4.4 Sep 2026 Official leaderboard
13 Kimi K3.1 Pro 44.4% ±4.2 Oct 2026 Official leaderboard
14 GPT-5.6 Sol 41.5% ±3.8 Jul 2026 Official leaderboard
15 Muse Wonder 39.1% ±3.8 Oct 2026 Official leaderboard
16 Claude Haiku 5.5 38.5% ±3.3 Oct 2026 Official leaderboard
17 Grok 4.6 38.1% ±3.2 Aug 2026 Official leaderboard
18 Gemini 3.8 Flash 35.8% ±3.2 Sep 2026 Official leaderboard
19 Grok 4.7 34.8% ±3.2 Sep 2026 Official leaderboard
20 Inkling 2 Small 32.7% ±2.5 Nov 2026 Official leaderboard
21 Ember-1 31.8% ±3.3 Sep 2026 Official leaderboard
22 MiMo-V2.6 Flash 31.7% ±2.5 Sep 2026 Official leaderboard
23 DeepSeek V4.1 Flash 29.6% ±1.9 Sep 2026 Official leaderboard
24 Step 5 Preview 29.6% ±2.2 Sep 2026 Official leaderboard
25 GPT-6 Luna 28.4% ±2.9 Sep 2026 Official leaderboard
26 Gemini 3.1 Pro 27.4% ±2.8 Feb 2026 Official leaderboard
27 GPT-5.6 Luna 25.3% ±1.6 Jul 2026 Official leaderboard
28 Beam 1.0 22% ±2.3 Oct 2026 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-6.2 Astra 68.4% Nov 2026 Official leaderboard
2 GPT-6 Astra 65.1% Sep 2026 Official leaderboard
3 Claude Fable 5.2 63.7% Nov 2026 Official leaderboard
4 Gemini 4 Neon 62.6% Nov 2026 Official leaderboard
5 GPT-6.1 Sol 59.6% Sep 2026 Official leaderboard
6 Claude Sonnet 5.5 59.1% Sep 2026 Official leaderboard
7 Claude Opus 5.5 58.6% Sep 2026 Official leaderboard
8 GPT-6.2 Sol 58.5% Nov 2026 Lab model card
9 Gemini 4 Argon 55.4% Sep 2026 Official leaderboard
10 Muse Wonder Full 48.5% Oct 2026 Official leaderboard
11 Claude Opus 5 47.9% Jul 2026 Official leaderboard
12 GPT-6 Sol 45% Sep 2026 Official leaderboard
13 Kimi K3.1 Pro 44.4% Oct 2026 Official leaderboard
14 GPT-5.6 Sol 41.5% Jul 2026 Official leaderboard
15 Muse Wonder 39.1% Oct 2026 Official leaderboard
16 Claude Haiku 5.5 38.5% Oct 2026 Official leaderboard
17 Grok 4.6 38.1% Aug 2026 Official leaderboard
18 Gemini 3.8 Flash 35.8% Sep 2026 Official leaderboard
19 Grok 4.7 34.8% Sep 2026 Official leaderboard
20 Inkling 2 Small 32.7% Nov 2026 Official leaderboard
21 Ember-1 31.8% Sep 2026 Official leaderboard
22 MiMo-V2.6 Flash 31.7% Sep 2026 Official leaderboard
23 DeepSeek V4.1 Flash 29.6% Sep 2026 Official leaderboard
24 Step 5 Preview 29.6% Sep 2026 Official leaderboard
25 GPT-6 Luna 28.4% Sep 2026 Official leaderboard
26 Gemini 3.1 Pro 27.4% Feb 2026 Official leaderboard
27 GPT-5.6 Luna 25.3% Jul 2026 Official leaderboard
28 Beam 1.0 22% Oct 2026 Official leaderboard

Labs report scores for 2 models in their cards. Switch the scores above to rank them together.

Accuracy against latency

The Vals scatter chart with its x axis switched from Cost to Latency: mean time per test against accuracy, on a log scale. The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

0% 20% 40% 60% 80% 1,0002,0005,00010K Latency (mean per test) (s), log scale, lower is better Accuracy GPT-6.2 AstraGPT-6.2 SolGemini 3.1 Pro 0% 20% 40% 60% 80% 1,0002,0005,00010K Latency (mean per test) (s), log scale, lower is better Accuracy GPT-6.2 AstraGPT-6.2 SolGemini 3.1 Pro
ModelSettingAccuracyLatency (mean per test)
GPT-6.2 Astra 68.4% 1,335 s
GPT-6 Astra 65.1% 2,437 s
Claude Fable 5.2 63.7% 2,081 s
Gemini 4 Neon 62.6% 2,452 s
GPT-6.1 Sol 59.6% 1,991 s
Claude Sonnet 5.5 59.1% 4,159 s
Claude Opus 5.5 58.7% 3,591 s
GPT-6.2 Sol 57.8% 1,140 s
Gemini 4 Argon 55.4% 3,420 s
Muse Wonder Full 48.5% 5,663 s
Claude Opus 5 47.9% 3,706 s
GPT-6 Sol 45% 1,876 s
Kimi K3.1 Pro 44.4% 1,382 s
GPT-5.6 Sol 41.5% 2,612 s
Muse Wonder 39.1% 2,588 s
Claude Haiku 5.5 38.5% 2,398 s
Grok 4.6 38.1% 2,410 s
Gemini 3.8 Flash 35.8% 1,320 s
Grok 4.7 34.8% 2,434 s
Inkling 2 Small 32.7% 3,449 s
Ember-1 31.8% 4,659 s
MiMo-V2.6 Flash 31.7% 6,048 s
DeepSeek V4.1 Flash 29.7% 3,999 s
Step 5 Preview 29.6% 4,018 s
GPT-6 Luna 28.4% 1,815 s
Gemini 3.1 Pro 27.4% 1,084 s
GPT-5.6 Luna 25.3% 3,000 s
Beam 1.0 22% 1,748 s

The best score, 68.4%, costs $4.76 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Xiaomi Corp OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% $0.10 $1.00 $10 $100 Cost per task, log scale GPT-6.2 Astra GPT-6.1 Sol MiMo-V2.6 Flash GPT-6 Luna 0% 20% 40% 60% 80% $0.10 $1.00 $10 $100 Cost per task, log scale GPT-6.2 Astra GPT-6.1 Sol

Source: Official Vals Science Index leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-6.2 Astra 68.4% $4.76 48,378
GPT-6 Astra 65.1% $8.03 65,321
Claude Fable 5.2 63.7% $8.79 94,053
Gemini 4 Neon 62.6% $8.26 220,508
GPT-6.1 Sol 59.6% $1.36 68,765
Claude Sonnet 5.5 59.1% $8.50 281,131
Claude Opus 5.5 58.6% $8.71 150,133
GPT-6.2 Sol 57.8% $1.12 47,721
Gemini 4 Argon 55.4% $18 196,916
Muse Wonder Full 48.5% $10 136,611
Claude Opus 5 47.9% $13
GPT-6 Sol 45% $2.31 61,916
Kimi K3.1 Pro 44.4% $3.33 71,899
GPT-5.6 Sol 41.5% $3.40
Muse Wonder 39.1% $8.74 137,061
Claude Haiku 5.5 38.5% $1.86 248,672
Grok 4.6 38.1% $2.45 85,489
Gemini 3.8 Flash 35.8% $2.98 94,594
Grok 4.7 34.8% $7.12 97,507
Inkling 2 Small 32.7% $1.99 208,107
Ember-1 31.8% $11 122,076
MiMo-V2.6 Flash 31.7% $0.12 157,496
DeepSeek V4.1 Flash 29.6% $0.29
Step 5 Preview 29.6% $1.78 170,536
GPT-6 Luna 28.4% $0.11 76,098
Gemini 3.1 Pro 27.4% $2.56 39,111
GPT-5.6 Luna 25.3% $0.26
Beam 1.0 22% 90,658

2 models have held number one on the official leaderboard since its release in October 2026 2 models have held number one with claims included since its release in October 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard 2 leaders

Source: Official Vals Science Index leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra GPT-6 Astra Claude Opus 5 GPT-5.6 Sol Gemini 3.1 Pro 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra Claude Opus 5

Sources: Official Vals Science Index leaderboard and labs’ model cards

One lab put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

OpenAI
DateLaunch postScoreAs the post writes it
Nov 17, 2026 GPT-6.2 Sol 58.5 Vals Science Index v1 (GPT-6.2 Sol Xhigh, 57.8 at Max)
Nov 17, 2026 GPT-6.2 Astra 68.4 Vals Science Index v1 (GPT-6.2 Astra Max)

2 lab claims next to the leaderboard

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 1 of them match, which usually means one side copied the other. Every benchmark.

Model Lab says Board says Gap
GPT-6.2 Sol 58.5% Vals Science Index v1 (GPT-6.2 Sol Xhigh, 57.8 at Max) 57.8% Leaderboard +0.7 pts
GPT-6.2 Astra 68.4% Vals Science Index v1 (GPT-6.2 Astra Max) 68.4% Leaderboard Same

More in Other

All 69 in Other
  • Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness. Claude Fable 5.2 64.6
  • Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
  • Decision Index Public v0.3 In use Decision models return probabilities over labeled options in public tasks covering knowledge, language, retrieval, tools, and arts. A chance-corrected weighted aggregate combines the task scores. Granite Switch 4.1 30B 51.3
  • SAGE v1.1 (Vals) In use Models grade handwritten undergraduate math and computer science work against rubrics. Scored by class-balanced accuracy of each rubric check against expert grading. Claude Opus 5.5 91.3%
  • MTEB Fading Massive Text Embedding Benchmark: retrieval, clustering, classification across languages and modalities. Qwen3-Embedding-8B 72.2%
  • SPEED-Bench Only in system cards Measures speculative decoding across varied prompts and domains. Reports the average acceptance length of draft tokens, so higher means faster generation at the same output. Qwen3-Next-80B-A3B-Instruct 2.81

Data audit

All 12 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 5 pass
  • Pass Independent runs Scores from the official leaderboard (28 entries) and launch posts (2 scores).
  • Pass Models scored 28 models we track.
  • Pass Newest score The newest is from November 2026, for Gemini 4 Neon.
  • Pass Frontier models 7 of the top 10 have a score, led by GPT-6.2 Astra (#1).
  • Pass Headroom The best is 68.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on vals.ai: 28 rows for models we track at the last read.
  • Pass Last read November 19th, 2026.
  • Pass Read worked The last read worked.
Consistency 2 pass
  • Pass Lab claims Both pairs agree within 10 points.
  • Pass Known issues No published audit found problems.

How we check