Benchmark Almanac

PerceptionBench (Moonshot)

GPT-5.6 Sol 59.7% leads since July 2026
Muse Ember 80.6% leads since November 2026
In use

The official leaderboard hasn’t scored a model released after Kimi K3 (July 2026). Labs report their own numbers for 5 newer models, such as Muse Ember at 80.6%.

Moonshot in-house benchmark of atomic visual perception skills. Scoring details unpublished.

Lab-created in-house eval (Moonshot AI). Not the same as other public benchmarks with this name.

Made by
Moonshot AI
Released
July 2026
Kind
General visual questions
Level
Everyday
Scored
Checked automatically
Answers
Short answer
Tasks
Kept private
Measures
Capability
In and out
Text, Images → Text
Interaction
Single turn
Items from
Written by experts
Refresh
One fixed set
Run by
A host
In views
Vision
Strongest model tested
Claude Fable 5 #12 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 18th, 2026
16
Models scored on its own leaderboard
6
Launch posts list it, from 4 labs
3
System cards report it, from 3 labs

GPT-5.6 Sol leads with 59.7% among models released by July 2026

Best score per model from the benchmark's own leaderboard, read on November 18th, 2026. Where a model ran with several settings or harnesses, we keep its best.

Muse Ember leads with 80.6%, 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.

The newest model with a score here came out in July 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.6 Sol 59.7%
  2. 2 Kimi K3 58.5%
  3. 3 Claude Fable 5 57.2%
  4. 4 Gemini 3.1 Pro 56.2%
  5. 5 GPT-5.5 55.8%
  6. 6 Seed2.1 Pro 55%
  7. 7 Gemini 3.5 Flash 52%
  8. 8 Qwen3.7-Plus 51.1%
  9. 9 Qwen3.5 397B-A17B 47.5%
  10. 10 Claude Opus 4.8 47.2%
  11. 11 Kimi K2.6 42.6%
  12. 12 Grok 4.5 41%
  13. 13 Gemma 4 31B IT 40.7%
  14. 14 GLM-5V-Turbo 39.6%
  15. 15 MiniMax-M3 33.1%
  16. 16 GLM-4.6V 32.5%

Source: Official PerceptionBench (Moonshot) leaderboard

#ModelScoreSettingDateSource
1 GPT-5.6 Sol 59.7% Jul 2026 Official leaderboard
2 Kimi K3 58.5% Jul 2026 Official leaderboard
3 Claude Fable 5 57.2% Jun 2026 Official leaderboard
4 Gemini 3.1 Pro 56.2% Feb 2026 Official leaderboard
5 GPT-5.5 55.8% Apr 2026 Official leaderboard
6 Seed2.1 Pro 55% Jun 2026 Official leaderboard
7 Gemini 3.5 Flash 52% May 2026 Official leaderboard
8 Qwen3.7-Plus 51.1% Jun 2026 Official leaderboard
9 Qwen3.5 397B-A17B 47.5% Feb 2026 Official leaderboard
10 Claude Opus 4.8 47.2% May 2026 Official leaderboard
11 Kimi K2.6 42.6% Apr 2026 Official leaderboard
12 Grok 4.5 41% Jul 2026 Official leaderboard
13 Gemma 4 31B IT 40.7% Apr 2026 Official leaderboard
14 GLM-5V-Turbo 39.6% Apr 2026 Official leaderboard
15 MiniMax-M3 33.1% Jun 2026 Official leaderboard
16 GLM-4.6V 32.5% Dec 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Muse Ember 80.6% max effort Nov 2026 Lab model card
2 Kimi K3.1 Pro 68.3% Oct 2026 Lab model card
3 Inkling 2 Small 64.7% Nov 2026 Lab model card
4 Qwen 3.8 Max 63.5% Aug 2026 Lab model card
5 GPT-5.6 Sol 59.7% Jul 2026 Official leaderboard
6 Kimi K3 58.5% Jul 2026 Official leaderboard
7 Inkling 2 Mini 57.5% Nov 2026 Lab model card
8 Claude Fable 5 57.2% Jun 2026 Official leaderboard
9 Gemini 3.1 Pro 56.2% Feb 2026 Official leaderboard
10 GPT-5.5 55.8% Apr 2026 Official leaderboard
11 Seed2.1 Pro 55% Jun 2026 Official leaderboard
12 Gemini 3.5 Flash 52% May 2026 Official leaderboard
13 Qwen3.7-Plus 51.1% Jun 2026 Official leaderboard
14 Qwen3.5 397B-A17B 47.5% Feb 2026 Official leaderboard
15 Claude Opus 4.8 47.2% May 2026 Official leaderboard
16 Kimi K2.6 42.6% Apr 2026 Official leaderboard
17 Grok 4.5 41% Jul 2026 Official leaderboard
18 Gemma 4 31B IT 40.7% Apr 2026 Official leaderboard
19 GLM-5V-Turbo 39.6% Apr 2026 Official leaderboard
20 MiniMax-M3 33.1% Jun 2026 Official leaderboard
21 GLM-4.6V 32.5% Dec 2025 Official leaderboard

Labs report scores for 6 models in their cards. Switch the scores above to rank them together.

Rankings by part

We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Depth, OCR and Visual relation. GPT-5.6 Sol leads 5 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Depth

16 models we track have a score here.

  1. 1 GPT-5.6 Sol 55.5%
  2. 2 Kimi K3 52.4%
  3. 3 Claude Fable 5 51.5%
  4. 4 Gemini 3.1 Pro 50%
  5. 5 Gemini 3.5 Flash ↑2 49.7%
  6. 6 GPT-5.5 ↓1 48.8%
  7. 7 Qwen3.7-Plus ↑1 48.5%
  8. 8 Qwen3.5 397B-A17B ↑1 44.6%
  9. 9 Seed2.1 Pro ↓3 43.6%
  10. 10 Kimi K2.6 ↑1 42.4%

2 models have held number one on the official leaderboard since its release in July 2026 5 models have held number one with claims included since its release in July 2026

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.

Anthropic OpenAI Moonshot AI Qwen Meta
Own leaderboard Not tested since Jul 2026
GPT-5.6 Sol 59.7%
Lab claims 4 leaders
Muse Ember 80.6%

Sources: Official PerceptionBench (Moonshot) leaderboard and labs’ model cards

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%) Muse Ember Kimi K3.1 Pro Qwen 3.8 Max Claude Fable 5 Gemini 3.1 Pro Qwen3.5 397B-A17B GLM-4.6V 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Muse Ember Kimi K3.1 Pro

Sources: Official PerceptionBench (Moonshot) leaderboard and labs’ model cards

4 labs 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.

Moonshot AI
Qwen
Thinking Machines
Meta
DateLaunch postScoreAs the post writes it
Nov 20, 2026 Inkling 2 Mini 57.5 PerceptionBench
Nov 9, 2026 Muse Ember 80.6 PerceptionBench with tools, Muse Ember max reasoning
Nov 5, 2026 Inkling 2 Small 64.7 PerceptionBench
Oct 24, 2026 Kimi K3.1 Pro 68.3 PerceptionBench
Aug 3, 2026 Qwen 3.8 Max 63.5 PerceptionBench
Jul 16, 2026 Kimi K3 58.5 PerceptionBench

3 system cards report it, from 3 labs

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 57.5 PerceptionBench Capabilities
Nov 9, 2026 Muse Ember technical report 80.6 PerceptionBench with tools, Muse Ember max reasoning Capabilities
Jul 27, 2026 Kimi K3 technical report 58.5 PerceptionBench (Table 2) 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
Kimi K3 58.5% PerceptionBench 58.5% Leaderboard Same

More in Vision

All 636 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. Qwen 3.6 Plus 86%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy. Seed2.1 Pro 90.7%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%

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 (16 entries), launch posts (6 scores) and system cards (3 scores).
  • Pass Models scored 21 models we track.
  • Pass Newest score The newest is from November 2026, for Inkling 2 Mini.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Fable 5 (#12).
  • Pass Headroom The best is 80.6%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 16 rows for models we track at the last read.
  • Pass Last read November 18th, 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