Benchmark Almanac

SAGE v1.1 (Vals)

Claude Opus 5.5 91.3% leads since October 2026
Claude Opus 5.5 91.3% leads since October 2026
In use
Version 1.1

The official leaderboard hasn’t scored a model released after Inkling 2 Small (November 2026). Labs report their own numbers for one newer model: Inkling 2 Mini at 85.9%.

Models grade handwritten undergraduate math and computer science work against rubrics. Scored by class-balanced accuracy of each rubric check against expert grading.

Vals changed the metric from v1, which moved strict graders up and lenient ones down. 545 tests in five suites. The board had 28 models on October 27th, 2026.

Made by
Vals AI
Released
October 2026
Kind
Other
Level
University
Scored
Graded by people
Answers
Open-ended
Tasks
Kept private
Measures
Capability
Language
English
In and out
Text, Documents → Text
Interaction
Single turn
Run by
A host
Based on
SAGE (Vals)
Strongest model tested
Claude Opus 5.5 #5 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 19th, 2026
29
Models scored on its own leaderboard
1
Launch post lists it, from 1 lab
1
System card reports it, from 1 lab

Claude Opus 5.5 leads with 91.3%

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.

Claude Opus 5.5 leads with 91.3%, 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 Claude Opus 5.5 91.3%
  2. 2 GPT-6.1 Sol 90.5%
  3. 3 Claude Haiku 5.5 89.8%
  4. 4 Gemini 4 Argon 89.5%
  5. 5 GLM-5.3-Flash 89.3%
  6. 6 Qwen3.7-Plus 89.2%
  7. 7 Grok 4.7 88.7%
  8. 8 Inkling 2 Small 88.6%
  9. 9 Gemini 3.8 Flash 88.2%
  10. 10 Claude Sonnet 5.5 88%
  11. 11 GPT-5.6 Terra 87%
  12. 12 Kimi K3 86.8%
  13. 13 Ember-1 86.4%
  14. 14 Muse Spark Open 85.8%
  15. 15 Qwen3.5 397B-A17B 85.5%
  16. 16 MiniMax-M3 85.5%
  17. 17 DeepSeek V4.1 Flash 85.4%
  18. 18 MiniMax-M3.1-Flash 85.3%
  19. 19 Qwen 3.5 Flash (hosted 35B-A3B) 84.9%
  20. 20 MiMo-V2.6 Pro 84.5%
  21. 21 GPT-6 Luna 83.4%
  22. 22 MiMo-V2.6 Flash 83.3%
  23. 23 Gemini 3.1 Pro 83.2%
  24. 24 Gemma 4 31B IT 82.8%
  25. 25 Qwen 3.8 Max 81.8%
  26. 26 Gemini 3.5 Flash-Lite 79.5%
  27. 27 Grok 4.1 Fast reasoning 74.5%
  28. 28 Mistral Medium 3.5 72%
  29. 29 Mistral Small 4 59.2%

Source: Official SAGE v1.1 (Vals) leaderboard

#ModelScore ± 95% CISettingDateSource
1 Claude Opus 5.5 91.3% ±2.7 Sep 2026 Official leaderboard
2 GPT-6.1 Sol 90.5% ±3 Sep 2026 Official leaderboard
3 Claude Haiku 5.5 89.8% ±2.8 Oct 2026 Official leaderboard
4 Gemini 4 Argon 89.5% ±3.2 Sep 2026 Official leaderboard
5 GLM-5.3-Flash 89.3% ±3.4 Aug 2026 Official leaderboard
6 Qwen3.7-Plus 89.2% ±2.7 Jun 2026 Official leaderboard
7 Grok 4.7 88.7% ±3.5 Sep 2026 Official leaderboard
8 Inkling 2 Small 88.6% ±3.4 Nov 2026 Official leaderboard
9 Gemini 3.8 Flash 88.2% ±3.4 Sep 2026 Official leaderboard
10 Claude Sonnet 5.5 88% ±3.2 Sep 2026 Official leaderboard
11 GPT-5.6 Terra 87% ±3.9 Jul 2026 Official leaderboard
12 Kimi K3 86.8% ±3.6 Jul 2026 Official leaderboard
13 Ember-1 86.4% ±3.7 Sep 2026 Official leaderboard
14 Muse Spark Open 85.8% ±3.4 Oct 2026 Official leaderboard
15 Qwen3.5 397B-A17B 85.5% ±3.6 Feb 2026 Official leaderboard
16 MiniMax-M3 85.5% ±3.4 Jun 2026 Official leaderboard
17 DeepSeek V4.1 Flash 85.4% ±3.8 Sep 2026 Official leaderboard
18 MiniMax-M3.1-Flash 85.3% ±3.9 Oct 2026 Official leaderboard
19 Qwen 3.5 Flash (hosted 35B-A3B) 84.9% ±3.6 Feb 2026 Official leaderboard
20 MiMo-V2.6 Pro 84.5% ±3.7 Sep 2026 Official leaderboard
21 GPT-6 Luna 83.4% ±4.5 Sep 2026 Official leaderboard
22 MiMo-V2.6 Flash 83.3% ±4.2 Sep 2026 Official leaderboard
23 Gemini 3.1 Pro 83.2% ±3.8 Feb 2026 Official leaderboard
24 Gemma 4 31B IT 82.8% ±4.1 Apr 2026 Official leaderboard
25 Qwen 3.8 Max 81.8% ±4.1 Aug 2026 Official leaderboard
26 Gemini 3.5 Flash-Lite 79.5% ±4 Jul 2026 Official leaderboard
27 Grok 4.1 Fast 74.5% ±4.4 reasoning Nov 2025 Official leaderboard
28 Mistral Medium 3.5 72% ±4.3 Apr 2026 Official leaderboard
29 Mistral Small 4 59.2% ±4.2 Mar 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Opus 5.5 91.3% Sep 2026 Official leaderboard
2 GPT-6.1 Sol 90.5% Sep 2026 Official leaderboard
3 Claude Haiku 5.5 89.8% Oct 2026 Official leaderboard
4 Gemini 4 Argon 89.5% Sep 2026 Official leaderboard
5 GLM-5.3-Flash 89.3% Aug 2026 Official leaderboard
6 Qwen3.7-Plus 89.2% Jun 2026 Official leaderboard
7 Grok 4.7 88.7% Sep 2026 Official leaderboard
8 Inkling 2 Small 88.6% Nov 2026 Lab model card
9 Gemini 3.8 Flash 88.2% Sep 2026 Official leaderboard
10 Claude Sonnet 5.5 88% Sep 2026 Official leaderboard
11 GPT-5.6 Terra 87% Jul 2026 Official leaderboard
12 Kimi K3 86.8% Jul 2026 Official leaderboard
13 Ember-1 86.4% Sep 2026 Official leaderboard
14 Inkling 2 Mini 85.9% Nov 2026 Lab model card
15 Muse Spark Open 85.8% Oct 2026 Official leaderboard
16 Qwen3.5 397B-A17B 85.5% Feb 2026 Official leaderboard
17 MiniMax-M3 85.5% Jun 2026 Official leaderboard
18 DeepSeek V4.1 Flash 85.4% Sep 2026 Official leaderboard
19 MiniMax-M3.1-Flash 85.3% Oct 2026 Official leaderboard
20 Qwen 3.5 Flash (hosted 35B-A3B) 84.9% Feb 2026 Official leaderboard
21 MiMo-V2.6 Pro 84.5% Sep 2026 Official leaderboard
22 GPT-6 Luna 83.4% Sep 2026 Official leaderboard
23 MiMo-V2.6 Flash 83.3% Sep 2026 Official leaderboard
24 Gemini 3.1 Pro 83.2% Feb 2026 Official leaderboard
25 Gemma 4 31B IT 82.8% Apr 2026 Official leaderboard
26 Qwen 3.8 Max 81.8% Aug 2026 Official leaderboard
27 Gemini 3.5 Flash-Lite 79.5% Jul 2026 Official leaderboard
28 Grok 4.1 Fast 74.5% reasoning Nov 2025 Official leaderboard
29 Mistral Medium 3.5 72% Apr 2026 Official leaderboard
30 Mistral Small 4 59.2% Mar 2026 Official leaderboard

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

The best score, 91.3%, costs $0.28 for the full run

Runs on the cost frontier score higher than any run that's cheaper. GLM-5.3-Flash comes within 2 points of the best for $0.006 for the full run, 44 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

xAI Qwen Mistral AI Z.ai Anthropic OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost of the full run, log scale Claude Opus 5.5 GPT-6.1 Sol Claude Haiku 5.5 GLM-5.3-Flash Qwen 3.5 Flash (hosted 35B-A3B) Grok 4.1 Fast, reasoning Mistral Small 4 0% 20% 40% 60% 80% 100% $0.001 $0.01 $0.10 $1.00 Cost of the full run, log scale Claude Opus 5.5 Claude Haiku 5.5

Source: Official SAGE v1.1 (Vals) leaderboard

ModelEffortHarnessScoreCost of the full runOutput tokens
Claude Opus 5.5 91.3% $0.28 10,157
GPT-6.1 Sol 90.5% $0.08 2,237
Claude Haiku 5.5 89.8% $0.02 28,718
Gemini 4 Argon 89.5% $0.07 2,533
GLM-5.3-Flash 89.3% $0.006 11,595
Qwen3.7-Plus 89.2% $0.02 6,561
Grok 4.7 88.7% $0.11 11,295
Inkling 2 Small 88.6% $0.02 10,882
Gemini 3.8 Flash 88.2% $0.08 9,377
Claude Sonnet 5.5 88% $0.16 12,190
GPT-5.6 Terra 87% $0.11 4,266
Kimi K3 86.8% $0.23 5,583
Ember-1 86.4% $0.21 4,053
Muse Spark Open 85.8% $0.07 4,090
Qwen3.5 397B-A17B 85.5% $0.01 8,922
MiniMax-M3 85.5% $0.01 5,435
DeepSeek V4.1 Flash 85.4% $0.01 7,114
MiniMax-M3.1-Flash 85.3% $0.008 5,865
Qwen 3.5 Flash (hosted 35B-A3B) 84.9% $0.006 8,945
MiMo-V2.6 Pro 84.5% $0.02 6,261
GPT-6 Luna 83.4% $0.006 5,652
MiMo-V2.6 Flash 83.3% $0.007 8,391
Gemini 3.1 Pro 83.2% $0.05 3,679
Gemma 4 31B IT 82.8% 3,349
Qwen 3.8 Max 81.8% $0.09 6,285
Gemini 3.5 Flash-Lite 79.5% $0.01 3,210
Grok 4.1 Fast reasoning 74.5% $0.005 1,100
Mistral Medium 3.5 72% $0.07 6,924
Mistral Small 4 59.2% $0.003 1,939

Number one over time Number one over time

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

Anthropic
Own leaderboard Passed 85% in Oct 2026, by Claude Opus 5.5

Source: Official SAGE v1.1 (Vals) leaderboard

Qwen3.5 397B-A17B on the official leaderboard passed 85% in February 2026

Models released before it came out already scored that high. 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%) Claude Opus 5.5 Qwen3.7-Plus Qwen3.5 397B-A17B Grok 4.1 Fast 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Opus 5.5 Grok 4.1 Fast

Sources: Official SAGE v1.1 (Vals) 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.

Thinking Machines
DateLaunch postScoreAs the post writes it
Nov 5, 2026 Inkling 2 Small 88.6 SAGE v1.1, 88.6%

One system card reports it

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Nov 20, 2026 Inkling 2 Mini model card 85.9 SAGE v1.1 Capabilities

One lab claim 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
Inkling 2 Small 88.6% SAGE v1.1, 88.6% 88.6% Leaderboard Same

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

Google DeepMind Anthropic
SAGE v1.1 (Vals) Passed 85% in Oct 2026, by Claude Opus 5.5
SAGE (Vals) 5 leaders
SAGE v1.1 (Vals) Passed 85% in Oct 2026, by Claude Opus 5.5

Source: Official leaderboards

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.
  • Vals Science Index In use Agents are scored on the equal average of BioMysteryBench, MysteryMechanism and Terminal-Bench-Science. GPT-6.2 Astra 68.4%
  • 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
  • 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 11 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (29 entries), launch posts (1 score) and system cards (1 score).
  • Pass Models scored 30 models we track.
  • Pass Newest score The newest is from November 2026, for Inkling 2 Mini.
  • Pass Frontier models 4 of the top 10 have a score, led by Claude Opus 5.5 (#5).
  • Note Headroom Saturated. The best is 91.3%, past 85% since February 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on vals.ai: 29 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 The one pair agrees within 10 points.
  • Pass Known issues No published audit found problems.

How we check