Leaderboard
Every model’s score, overall and test by test. Open a model for its full breakdown.
- #1Claude Opus 5.5Anthropic79.2%221.6 / 280 pts
- #2Claude Fable 5.1Anthropic75.3%210.8 / 280 pts
- #3GPT-6 AstraOpenAI71.3%199.7 / 280 pts
Rankings
Share of the 280 available points, across all 5 tests.
- Top score
- Other models
- 1Claude Opus 5.5Top score in 4 of 5 tests79.2% · 221.6
- 2Claude Fable 5.175.3% · 210.8
- 3GPT-6 Astra71.3% · 199.7
- 4Nex N2.5 Pro69.6% · 194.9
- 5Kimi K3Top in Cycles68.3% · 191.4
- 6Muse Spark 1.367.8% · 190.0
- 7Grok 4.7Top in Placement67.1% · 187.8
- 8Gemini 3.7 Flash61.1% · 171.1
- 9GPT-5.6 Sol59.6% · 166.8
- 10Muse Spark 1.2Top in Cycles57.2% · 160.2
- 11Qwen 3.8 Max56.2% · 157.3
- 12Gemini 3.1 Pro53.2% · 149.0
- 13DeepSeek V4 Pro48.8% · 136.7
0%25%50%75%100%
By test
Each test ranked on its own, heaviest first. Bars run to that test's maximum.
- Top score in that test
- Other models
Micro-contents
out of 100 · 36% of total- Claude Opus 5.577.9
- Claude Fable 5.176.3
- GPT-6 Astra74.5
- Nex N2.5 Pro65.8
- Grok 4.765.0
- Muse Spark 1.364.9
- Kimi K360.1
- GPT-5.6 Sol57.5
- Muse Spark 1.255.3
- Gemini 3.7 Flash52.2
- Qwen 3.8 Max48.0
- Gemini 3.1 Pro44.3
- DeepSeek V4 Pro38.0
Physics
out of 100 · 36% of total- Claude Opus 5.574.8
- Claude Fable 5.172.5
- Nex N2.5 Pro63.8
- GPT-6 Astra63.3
- Kimi K362.3
- Muse Spark 1.361.8
- GPT-5.6 Sol59.3
- Grok 4.758.8
- Qwen 3.8 Max57.3
- Gemini 3.7 Flash56.8
- Muse Spark 1.250.5
- Gemini 3.1 Pro44.8
- DeepSeek V4 Pro42.3
Voxel island
out of 40 · 14% of total- Claude Opus 5.540.0
- Kimi K338.0
- Nex N2.5 Pro37.3
- Muse Spark 1.337.3
- Gemini 3.7 Flash37.1
- Gemini 3.1 Pro37.0
- Claude Fable 5.136.0
- GPT-6 Astra36.0
- Grok 4.735.0
- DeepSeek V4 Pro30.5
- Muse Spark 1.226.4
- GPT-5.6 Sol26.0
- Qwen 3.8 Max26.0
Coverage and placement
out of 20 · 7% of totalTemporal cycles
out of 20 · 7% of totalWhere the points come from
Each model's total, split into its tests.
- Micro-contents (100)
- Physics (100)
- Voxel island (40)
- Coverage and placement (20)
- Temporal cycles (20)
- Claude Opus 5.5221.6
- Claude Fable 5.1210.8
- GPT-6 Astra199.7
- Nex N2.5 Pro194.9
- Kimi K3191.4
- Muse Spark 1.3190.0
- Grok 4.7187.8
- Gemini 3.7 Flash171.1
- GPT-5.6 Sol166.8
- Muse Spark 1.2160.2
- Qwen 3.8 Max157.3
- Gemini 3.1 Pro149.0
- DeepSeek V4 Pro136.7
All scores
Bold values are the top score in their column.
| # | Model | Total | Contents /100 | Physics /100 | Voxel /40 | Placement /20 | Cycles /20 |
|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 221.6 / 280 | 77.9 | 74.8 | 40.0 | 18 | 11 |
| 2 | Claude Fable 5.1 | 210.8 / 280 | 76.3 | 72.5 | 36.0 | 15 | 11 |
| 3 | GPT-6 Astra | 199.7 / 280 | 74.5 | 63.3 | 36.0 | 16 | 10 |
| 4 | Nex N2.5 Pro | 194.9 / 280 | 65.8 | 63.8 | 37.3 | 15 | 13 |
| 5 | Kimi K3 | 191.4 / 280 | 60.1 | 62.3 | 38.0 | 17 | 14 |
| 6 | Muse Spark 1.3 | 190.0 / 280 | 64.9 | 61.8 | 37.3 | 15 | 11 |
| 7 | Grok 4.7 | 187.8 / 280 | 65.0 | 58.8 | 35.0 | 18 | 11 |
| 8 | Gemini 3.7 Flash | 171.1 / 280 | 52.2 | 56.8 | 37.1 | 14 | 11 |
| 9 | GPT-5.6 Sol | 166.8 / 280 | 57.5 | 59.3 | 26.0 | 13 | 11 |
| 10 | Muse Spark 1.2 | 160.2 / 280 | 55.3 | 50.5 | 26.4 | 14 | 14 |
| 11 | Qwen 3.8 Max | 157.3 / 280 | 48.0 | 57.3 | 26.0 | 14 | 12 |
| 12 | Gemini 3.1 Pro | 149.0 / 280 | 44.3 | 44.8 | 37.0 | 13 | 10 |
| 13 | DeepSeek V4 Pro | 136.7 / 280 | 38.0 | 42.3 | 30.5 | 14 | 12 |