Agents · Benchmark profile
A game-environment benchmark that evaluates AI models in novel games covering strategic planning, resource management, spatial reasoning, cooperation, and theory of mind.
Data verified 27 Jul 2026 · Methodology 1.8.0
Benchmark score on Gert Labs Composite Game Benchmark
Claude Opus 4.8 leads at 73.0%, followed by GPT-5.5 (72.9%) and Claude Opus 4.7 (65.6%).
10 ranked models · higher is better · labels show rank and score
| Model | Rank | Provider | Score |
|---|---|---|---|
| Claude Opus 4.8 | #1 | Anthropic | 73.0% |
| GPT-5.5 | #2 | OpenAI | 72.9% |
| Claude Opus 4.7 | #3 | Anthropic | 65.6% |
| GPT-5.4 | #4 | OpenAI | 64.9% |
| Qwen3.7-Max | #5 | Alibaba Cloud | 64.3% |
| Claude Opus 4.5 | #6 | Anthropic | 64.2% |
| Gemini 3 Pro | #7 | 63.2% | |
| Claude Sonnet 4.6 | #8 | Anthropic | 62.9% |
| MiMo-V2.5-Pro | #9 | Xiaomi | 62.7% |
| Claude Opus 4.6 | #10 | Anthropic | 61.9% |
One best score per model · higher is better
| Rank | Model | Provider | License | Evidence use | Score |
|---|---|---|---|---|---|
| #1 | Claude Opus 4.8 claude-opus-4-8 | Anthropic | closed | Reference only | 73.0% |
| #2 | GPT-5.5 gpt-5.5 | OpenAI | closed | Estimated reference | 72.9% |
| #3 | Claude Opus 4.7 claude-opus-4-7 | Anthropic | closed | Reference only | 65.6% |
| #4 | GPT-5.4 gpt-5.4 | OpenAI | closed | Reference only | 64.9% |
| #5 | Qwen3.7-Max qwen3.7-max | Alibaba Cloud | closed | Reference only | 64.3% |
| #6 | Claude Opus 4.5 claude-opus-4-5 | Anthropic | closed | Reference only | 64.2% |
| #7 | Gemini 3 Pro gemini-3-pro | closed | Reference only | 63.2% | |
| #8 | Claude Sonnet 4.6 claude-sonnet-4-6 | Anthropic | closed | Reference only | 62.9% |
| #9 | MiMo-V2.5-Pro mimo-v2-5-pro | Xiaomi | closed | Reference only | 62.7% |
| #10 | Claude Opus 4.6 claude-opus-4-6 | Anthropic | closed | Reference only | 61.9% |
| #11 | Gemini 3.5 Flash gemini-3.5-flash | closed | Estimated reference | 61.9% | |
| #12 | GLM-5.1 glm-5-1 | Z.AI | open | Reference only | 60.1% |
| #13 | GPT-5.3-Codex gpt-5.3-codex | OpenAI | closed | Reference only | 57.5% |
| #14 | Gemini 3.1 Pro gemini-3-1-pro | closed | Estimated reference | 56.9% | |
| #15 | Kimi K2.6 kimi-k2.6 | Moonshot AI | open | Estimated reference | 56.8% |
| #16 | Gemini 3 Flash gemini-3-flash | closed | Reference only | 56.6% | |
| #17 | Qwen3.6 27B qwen3.6-27b | Alibaba Cloud | open | Estimated reference | 54.8% |
| #18 | DeepSeek V4 Flash deepseek-v4-flash | DeepSeek | open | Estimated reference | 54.4% |
| #19 | GPT-5.2 Codex gpt-5.2-codex | OpenAI | closed | Reference only | 51.8% |
| #20 | Step 3.7 Flash step-3-7-flash | StepFun | open | Estimated reference | 51.6% |
| #21 | GLM-5 glm-5 | Z.AI | open | Reference only | 51.0% |
| #22 | Qwen3.6 Plus qwen3-6-plus | Alibaba Cloud | closed | Reference only | 50.6% |
| #23 | DeepSeek V4 Pro deepseek-v4-pro | DeepSeek | open | Reference only | 50.3% |
| #24 | GPT-5.1-Codex gpt-5-1-codex | OpenAI | closed | Estimated reference | 49.7% |
| #25 | Grok Build 0.1 grok-build-0-1 | xAI | closed | Estimated reference | 49.1% |
| #26 | Claude Sonnet 4.5 claude-sonnet-4-5 | Anthropic | closed | Estimated reference | 48.5% |
| #27 | Grok 4.1 Fast grok-4-1-fast | xAI | closed | Reference only | 47.3% |
| #28 | MiMo-V2.5 mimo-v2.5 | Xiaomi | open | Estimated reference | 46.9% |
| #29 | Qwen3.5 397B qwen3.5-397b-a17b | Alibaba Cloud | open | Estimated reference | 46.8% |
| #30 | GPT-5.2 gpt-5.2 | OpenAI | closed | Estimated reference | 46.5% |
| #31 | Kimi K2.5 kimi-k2.5 | Moonshot AI | open | Reference only | 45.9% |
| #32 | Grok 4.3 grok-4.3 | xAI | closed | Reference only | 43.9% |
| #33 | Qwen3 Max qwen3-max | Alibaba Cloud | closed | Estimated reference | 43.7% |
| #34 | Qwen3.6-35B-A3B qwen3-6-35b-a3b | Alibaba Cloud | open | Estimated reference | 42.6% |
| #35 | Grok 4 grok-4 | xAI | closed | Reference only | 42.3% |
| #36 | Gemini 2.5 Pro gemini-2.5-pro | closed | Reference only | 42.0% | |
| #37 | GPT-5.1 gpt.5.1 | OpenAI | closed | Estimated reference | 41.2% |
| #38 | MiniMax M2.7 minimax-m2-7 | MiniMax | open | Reference only | 40.4% |
| #39 | GLM-4.7 glm-4.7 | Z.AI | open | Reference only | 40.0% |
| #40 | Claude 4 Sonnet claude-4-sonnet | Anthropic | closed | Reference only | 39.7% |
| #41 | Qwen3.5 27B qwen3.5-27b | Alibaba Cloud | open | Reference only | 39.4% |
| #42 | Mistral Medium 3.5 128B mistral-medium-3-5-128b | Mistral AI | open | Estimated reference | 39.1% |
| #43 | Gemini 3.1 Flash-Lite gemini-3.1-flash-lite | closed | Reference only | 38.5% | |
| #44 | Grok 4.20 grok-4.20 | xAI | closed | Estimated reference | 38.4% |
| #45 | Hy3 Preview hy3-preview | Tencent | open | Estimated reference | 36.9% |
| #46 | MiMo-V2-Pro mimo-v2-pro | Xiaomi | closed | Reference only | 36.7% |
| #47 | Gemma 4 31B gemma-4-31b | open | Reference only | 35.3% | |
| #48 | Kimi K2.5 (Reasoning) kimi-k2-5-reasoning | Moonshot AI | closed | Estimated reference | 32.6% |
| #49 | Trinity-Large-Thinking trinity-large-thinking | Arcee AI | open | Reference only | 32.5% |
| #50 | GLM-5V-Turbo glm-5v-turbo | Z.AI | closed | Reference only | 30.8% |
| #51 | GPT-OSS 120B gpt-oss-120b | OpenAI | open | Reference only | 29.6% |
| #52 | DeepSeek V3.2 deepseek-v3.2 | DeepSeek | open | Reference only | 29.6% |
| #53 | Qwen3.5-35B-A3B qwen3-5-35b-a3b | Alibaba Cloud | open | Reference only | 29.0% |
| #54 | GPT-4.1 gpt-4-1 | OpenAI | closed | Reference only | 25.6% |
The top of this snapshot is led by Claude Opus 4.8 at 73.0%; third place is 7.4 points behind. The top-10 spread is 11.1 points.
About Gert Labs Composite Game Benchmark
A game-environment benchmark that evaluates AI models in novel games covering strategic planning, resource management, spatial reasoning, cooperation, and theory of mind. Results stay tied to the exact model variant and evaluation system. Multiple systems for the same model use the best published score on this page; overall Lumina scoring uses the median of ranking-eligible rows.
Open benchmark source ↗FAQ
A game-environment benchmark that evaluates AI models in novel games covering strategic planning, resource management, spatial reasoning, cooperation, and theory of mind.
Claude Opus 4.8 by Anthropic currently leads with 73.0%.
54 models in the LuminaBench cohort have a qualifying score on this benchmark.
No. This benchmark is display-only and does not enter the overall Lumina composite.
Related
