Opus Magnum Bench
Agents build machines on a hexagonal grid to make a target molecule. Each puzzle score compares the machine’s combined cost, cycles, and area with the human record.
The makers compare 20 models on 36 campaign puzzles. Unsolved puzzles earn zero. Matching the human record earns full credit.
- Made by
- Rob Haisfield, Max Bittker, and Surya Narreddi
- Released
- June 2026
- Kind
- Puzzles and chess
- Level
- Competition
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- Opus Magnum
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- Claude Fable 5 #12 on the capability index
- Last model added
- July 2026
- Scores refreshed
- November 20th, 2026
GPT-5.6 Sol leads with 25.5 among models released by July 2026
Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.
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:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 13 more
- 25.5 GPT-5.6 Sol
- 21.6 Claude Fable 5
- 19.4 GPT-5.5
- 17.2 GPT-5.6 Terra
- 11.2
Gemini 3.1 Pro
- 9.56 Claude Sonnet 5
- 9.14 GLM-5.2
- 8.82
Gemini 3.5 Flash
- 8.21 Claude Opus 4.8
- 6.55 Grok 4.5
- 6.41 GPT-5.6 Luna
- 5.54
Gemini 3 Flash
- 4.82
DeepSeek-V4-Pro
- 4.45 Claude Sonnet 4.6
- 3.93 Kimi K2.7 Code
- 3.88 GLM-5.1
- 3.32 GPT-5.4 Mini
- 2.35 GPT-5.4 Nano
- 1.98 Kimi K2.6
- 0.00 Grok 4.3
- 1 GPT-5.6 Sol xhigh 25.5
- 2 Claude Fable 5 high 21.6
- 3 GPT-5.5 xhigh 19.4
- 4 GPT-5.6 Terra xhigh 17.2
- 5
Gemini 3.1 Pro high 11.2
- 6 Claude Sonnet 5 high 9.56
- 7 GLM-5.2 high 9.14
- 8
Gemini 3.5 Flash high 8.82
- 9 Claude Opus 4.8 high 8.21
- 10 Grok 4.5 high 6.55
- 11 GPT-5.6 Luna xhigh 6.41
- 12
Gemini 3 Flash high 5.54
- 13
DeepSeek-V4-Pro 4.82
- 14 Claude Sonnet 4.6 medium 4.45
- 15 Kimi K2.7 Code 3.93
- 16 GLM-5.1 max 3.88
- 17 GPT-5.4 Mini xhigh 3.32
- 18 GPT-5.4 Nano xhigh 2.35
- 19 Kimi K2.6 1.98
- 20 Grok 4.3 high 0.00
Source: Official Opus Magnum Bench leaderboard
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.6 Sol | 25.5 | xhigh | Jul 2026 | Official leaderboard |
| 2 | Claude Fable 5 | 21.6 | high | Jun 2026 | Official leaderboard |
| 3 | GPT-5.5 | 19.4 | xhigh | Apr 2026 | Official leaderboard |
| 4 | GPT-5.6 Terra | 17.2 | xhigh | Jul 2026 | Official leaderboard |
| 5 | | 11.2 | high | Feb 2026 | Official leaderboard |
| 6 | Claude Sonnet 5 | 9.56 | high | Jun 2026 | Official leaderboard |
| 7 | GLM-5.2 | 9.14 | high | Jun 2026 | Official leaderboard |
| 8 | | 8.82 | high | May 2026 | Official leaderboard |
| 9 | Claude Opus 4.8 | 8.21 | high | May 2026 | Official leaderboard |
| 10 | Grok 4.5 | 6.55 | high | Jul 2026 | Official leaderboard |
| 11 | GPT-5.6 Luna | 6.41 | xhigh | Jul 2026 | Official leaderboard |
| 12 | | 5.54 | high | Dec 2025 | Official leaderboard |
| 13 | | 4.82 | Apr 2026 | Official leaderboard | |
| 14 | Claude Sonnet 4.6 | 4.45 | medium | Feb 2026 | Official leaderboard |
| 15 | Kimi K2.7 Code | 3.93 | Jun 2026 | Official leaderboard | |
| 16 | GLM-5.1 | 3.88 | max | Apr 2026 | Official leaderboard |
| 17 | GPT-5.4 Mini | 3.32 | xhigh | Mar 2026 | Official leaderboard |
| 18 | GPT-5.4 Nano | 2.35 | xhigh | Mar 2026 | Official leaderboard |
| 19 | Kimi K2.6 | 1.98 | Apr 2026 | Official leaderboard | |
| 20 | Grok 4.3 | 0.00 | high | Apr 2026 | Official leaderboard |
Rankings by part
We read 6 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Reward (avg), Solved (pass@k) and Avg steps. GPT-5.6 Terra leads 3 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Reward (avg)
20 models we track have a score here.
- 1 GPT-5.6 Sol 21.5
- 2 Claude Fable 5 17.0
- 3 GPT-5.5 14.8
- 4 GPT-5.6 Terra 12.2
- 5
Gemini 3.1 Pro 8.16
- 6 Claude Sonnet 5 7.35
- 7 GLM-5.2 6.43
- 8
Gemini 3.5 Flash 5.57
- 9 Claude Opus 4.8 5.56
- 10 GPT-5.6 Luna ↑1 4.62
Solved (pass@k)
20 models we track have a score here.
- 1 GPT-5.6 Sol 35.0
- 2 GPT-5.5 ↑1 31.0
- 3 Claude Fable 5 ↓1 28.0
- 4 GPT-5.6 Terra 28.0
- 5
Gemini 3.1 Pro 18.0
- 6 Claude Sonnet 5 17.0
- 7 GLM-5.2 15.0
- 8
Gemini 3.5 Flash 14.0
- 9 GPT-5.6 Luna ↑2 12.0
- 10 Claude Opus 4.8 ↓1 11.0
Avg steps
20 models we track have a score here.
- 1 GPT-5.6 Terra ↑3 119.8
- 2 GPT-5.6 Sol ↓1 117.8
- 3 Claude Sonnet 5 ↑3 115.3
- 4
DeepSeek-V4-Pro ↑9 114.1
- 5 GLM-5.2 ↑2 114.1
- 6
Gemini 3.5 Flash ↑2 113.7
- 7 GPT-5.6 Luna ↑4 109.6
- 8
Gemini 3 Flash ↑4 109.5
- 9 Claude Opus 4.8 102.4
- 10 Grok 4.3 ↑10 98.2
Steps to first solve
19 models we track have a score here.
- 1
DeepSeek-V4-Pro ↑12 59.9
- 2
Gemini 3.5 Flash ↑6 56.7
- 3 Claude Sonnet 5 ↑3 56.5
- 4 Kimi K2.7 Code ↑11 52.4
- 5 GLM-5.2 ↑2 49.8
- 6 GPT-5.6 Terra ↓2 46.3
- 7 Claude Opus 4.8 ↑2 44.2
- 8 GPT-5.4 Nano ↑10 37.3
- 9 GPT-5.6 Luna ↑2 34.7
- 10 GPT-5.5 ↓7 33.3
Steps to final solve
19 models we track have a score here.
- 1 GPT-5.6 Terra ↑3 99.7
- 2 GPT-5.6 Sol ↓1 94.2
- 3 GLM-5.2 ↑4 81.7
- 4 Claude Sonnet 5 ↑2 78.3
- 5
DeepSeek-V4-Pro ↑8 75.5
- 6
Gemini 3.5 Flash ↑2 75.3
- 7 GPT-5.6 Luna ↑4 74.6
- 8 Kimi K2.7 Code ↑7 67.6
- 9 Claude Opus 4.8 64.3
- 10 GPT-5.5 ↓7 62.1
Steps on fail
20 models we track have a score here.
- 1 GPT-5.6 Terra ↑3 120.0
- 2 Claude Sonnet 5 ↑4 117.3
- 3
Gemini 3.5 Flash ↑5 115.7
- 4
DeepSeek-V4-Pro ↑9 114.1
- 5 GPT-5.6 Luna ↑6 113.7
- 6 GLM-5.2 ↑1 113.2
- 7 GPT-5.6 Sol ↓6 110.7
- 8
Gemini 3 Flash ↑4 109.8
- 9 Claude Opus 4.8 98.9
- 10 Grok 4.3 ↑10 98.2
More effort adds a median 6 score
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: Official Opus Magnum Bench leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-5.6 Sol | xhigh | 25.5 | 385,225 | ||
| Claude Fable 5 | high | 21.6 | 756,266 | ||
| GPT-5.5 | xhigh | 19.4 | 244,835 | ||
| Claude Fable 5 | low | 17.7 | 358,886 | ||
| GPT-5.6 Terra | xhigh | 17.2 | 382,336 | ||
| GPT-5.5 | medium | 13.6 | 102,698 | ||
| high | 11.2 | 89,945 | |||
| Claude Sonnet 5 | high | 9.56 | 228,239 | ||
| GLM-5.2 | high | 9.14 | 592,156 | ||
| high | 8.82 | 121,764 | |||
| Claude Opus 4.8 | high | 8.21 | 40,493 | ||
| Grok 4.5 | high | 6.55 | 479,775 | ||
| GPT-5.6 Luna | xhigh | 6.41 | 788,601 | ||
| high | 5.54 | 172,286 | |||
| 4.82 | 407,001 | ||||
| Claude Opus 4.8 | 4.56 | 29,321 | |||
| Claude Sonnet 4.6 | medium | 4.45 | 23,035 | ||
| Kimi K2.7 Code | 3.93 | 172,793 | |||
| GLM-5.1 | max | 3.88 | 283,971 | ||
| GPT-5.4 Mini | xhigh | 3.32 | 999,663 | ||
| GPT-5.4 Nano | xhigh | 2.35 | 959,796 | ||
| Kimi K2.6 | 1.98 | 342,876 | |||
| Grok 4.3 | high | 0.00 | 56,741 |
3 models have held number one on the official leaderboard since its release in June 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official Opus Magnum Bench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official Opus Magnum Bench leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Games and long tasks
- ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6.2 Astra 100%
- Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games.
Gemini 2.5 Pro 3.90
- MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks.
Seed2.0 Pro 49%
- TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress.
Llama 4 Scout 7.00
- GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
- Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%
Data audit
All 9 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (20 entries).
- Pass Models scored 20 models we track.
- Pass Newest score The newest is from July 2026, for GPT-5.6 Luna.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude Fable 5 (#12).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on opusmagnumbench.com: 23 rows for models we track at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.