SimpleBench
Trick questions about space, time and social reasoning.
Agent Index weight
0%
shown for context only
Models tested
100
from the benchmark’s leaderboard and reports
Results
| # | Model | Score | Normalised | Harness | Cost/task | Date |
|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.2 Supported Anthropic | 94.9% | source | — | — | |
| 2 | Gemini 4 Carbon Supported Google | 92.6% | source | — | — | |
| 3 | Gemini 4 Argon Supported Google | 91.7% | source | — | — | |
| 4 | Claude Opus 5.5 Supported Anthropic | 88.4% | source | — | — | |
| 5 | Claude Fable 5.1 Supported Anthropic | 86.6% | source | — | — | |
| 6 | GPT-6 Astra Supported OpenAI | 86.5% | source | — | — | |
| 7 | GPT-6.1 Sol Estimated OpenAI | 82.9% | source | — | — | |
| 8 | Gemini 3.8 Flash Supported Google | 82.4% | source | — | — | |
| 9 | Claude Fable 5 Supported Anthropic | 81.9% | source | — | — | |
| 10 | Muse Spark 1.3 Estimated Meta | 81.8% | source | — | — | |
| 11 | Claude Opus 5 Supported Anthropic | 80.6% | source | — | — | |
| 12 | Gemini 3.1 Pro Preview Supported Google | 79.6% | source | — | — | |
| 13 | GPT-5.5 Pro OpenAI | 76.9% | source | — | — | |
| 14 | Gemini 3.5 Flash Supported Google | 76.7% | source | — | — | |
| 15 | Gemini 3 Pro Preview Supported Google | 76.4% | source | — | — | |
| 16 | Claude Sonnet 5.5 Estimated Anthropic | 75.9% | source | — | — | |
| 17 | Grok 4.6 Supported xAI | 75.9% | source | — | — | |
| 18 | Muse Spark 1.2 Estimated Meta | 74.5% | source | — | — | |
| 19 | GPT-5.4 Pro OpenAI | 74.1% | source | — | — | |
| 20 | GPT-6 Sol Estimated OpenAI | 73.1% | source | — | — | |
| 21 | Muse Ember Meta | 72.7% | source | — | — | |
| 22 | GPT-6.1 Terra Estimated OpenAI | 72.2% | source | — | — | |
| 23 | GPT-5.6 Sol Supported OpenAI | 71.7% | source | — | — | |
| 24 | Qwen3.7 Max Alibaba (Qwen) | 70.4% | source | — | — | |
| 25 | Grok 4.5 Estimated xAI | 70.0% | source | — | — | |
| 26 | GPT-5.5 Supported OpenAI | 69.0% | source | — | — | |
| 27 | Claude Opus 4.6 Supported Anthropic | 67.6% | source | — | — | |
| 28 | DeepSeek V4.1 Flash Estimated DeepSeek | 66.7% | source | — | — | |
| 29 | GLM-5.3 Supported Z.ai (Zhipu) | 66.2% | source | — | — | |
| 30 | Claude Opus 4.8 Supported Anthropic | 64.8% | source | — | — | |
| 31 | Qwen3.6 Max Preview Estimated Alibaba (Qwen) | 63.0% | source | — | — | |
| 32 | Gemini 2.5 Pro Supported Google | 62.4% | source | — | — | |
| 33 | Claude Opus 4.5 (latest) Supported Anthropic | 62.0% | source | — | — | |
| 34 | Claude Opus 4.7 Supported Anthropic | 61.7% | source | — | — | |
| 35 | GPT-5 Pro OpenAI | 61.6% | source | — | — | |
| 36 | DeepSeek V4 Flash 0731 DeepSeek | 61.1% | source | — | — | |
| 37 | Gemini 3 Flash Preview Supported Google | 61.1% | source | — | — | |
| 38 | Kimi K3 Supported Moonshot AI | 60.7% | source | — | — | |
| 39 | Claude Sonnet 5 Supported Anthropic | 60.6% | source | — | — | |
| 40 | Grok 4 Supported xAI | 60.5% | source | — | — | |
| 41 | Qwen3.8 27B Alibaba (Qwen) | 60.2% | source | — | — | |
| 42 | Claude Opus 4.1 (latest) Supported Anthropic | 60.0% | source | — | — | |
| 43 | Claude Haiku 5.5 Estimated Anthropic | 59.2% | source | — | — | |
| 44 | Claude-Opus-4 Supported Anthropic | 58.8% | source | — | — | |
| 45 | GLM-5.2 Supported Z.ai (Zhipu) | 58.8% | source | — | — | |
| 46 | Kimi K2.7 Code Estimated Moonshot AI | 57.9% | source | — | — | |
| 47 | GPT-5.2 Pro OpenAI | 57.4% | source | — | — | |
| 48 | GPT-5 Supported OpenAI | 56.7% | source | — | — | |
| 49 | Grok 4.1 Fast (Non-Reasoning) xAI | 56.0% | source | — | — | |
| 50 | GLM-5.1 Estimated Z.ai (Zhipu) | 55.1% | source | — | — | |
| 51 | Claude Sonnet 4.5 (latest) Supported Anthropic | 54.3% | source | — | — | |
| 52 | GLM-5 Supported Z.ai (Zhipu) | 53.2% | source | — | — | |
| 53 | GPT-5.1 Supported OpenAI | 53.2% | source | — | — | |
| 54 | o3 Supported OpenAI | 53.1% | source | — | — | |
| 55 | DeepSeek-V3.2-Speciale DeepSeek | 52.6% | source | — | — | |
| 56 | Inkling Estimated thinkingmachines | 50.0% | source | — | — | |
| 57 | GPT-5.6 Terra Estimated OpenAI | 48.9% | source | — | — | |
| 58 | GLM-4.7 Estimated Z.ai (Zhipu) | 47.7% | source | — | — | |
| 59 | GPT-5.6 Luna Estimated OpenAI | 46.8% | source | — | — | |
| 60 | Kimi K2.5 Supported Moonshot AI | 46.8% | source | — | — | |
| 61 | Claude 3.7 Sonnet Supported Anthropic | 46.4% | source | — | — | |
| 62 | DeepSeek V4 Flash DeepSeek | 46.3% | source | — | — | |
| 63 | GPT-5.2 Supported OpenAI | 45.8% | source | — | — | |
| 64 | MiniMax-M3 Estimated MiniMax | 45.8% | source | — | — | |
| 65 | Claude Sonnet 4 Supported Anthropic | 45.5% | source | — | — | |
| 66 | Nemotron 3 Ultra NVIDIA | 41.7% | source | — | — | |
| 67 | o1 Estimated OpenAI | 41.7% | source | — | — | |
| 68 | Claude-Sonnet-3.5 Estimated Anthropic | 41.4% | source | — | — | |
| 69 | Gemini 2.5 Flash Estimated Google | 41.2% | source | — | — | |
| 70 | DeepSeek-R1 Estimated DeepSeek | 40.8% | source | — | — | |
| 71 | DeepSeek V3.1 DeepSeek | 40.0% | source | — | — | |
| 72 | Kimi K2 Thinking Estimated Moonshot AI | 39.6% | source | — | — | |
| 73 | o4-mini Estimated OpenAI | 38.7% | source | — | — | |
| 74 | Grok 3 xAI | 36.1% | source | — | — | |
| 75 | Qwen3.6 Flash Alibaba (Qwen) | 35.2% | source | — | — | |
| 76 | MiniMax-M2.1 MiniMax | 34.7% | source | — | — | |
| 77 | GPT-4.5 OpenAI | 34.5% | source | — | — | |
| 78 | Gemini-2.0-Flash Google | 31.1% | source | — | — | |
| 79 | Qwen3 235B A22B Alibaba (Qwen) | 31.0% | source | — | — | |
| 80 | Gemini 2.0 Flash Thinking (Jan 2025) Google | 30.7% | source | — | — | |
| 81 | Llama 4 Maverick Meta | 27.7% | source | — | — | |
| 82 | DeepSeek V3 0324 DeepSeek | 27.2% | source | — | — | |
| 83 | Gemini 1.5 Pro (Feb 2024) Google | 27.1% | source | — | — | |
| 84 | GPT-4.1 OpenAI | 27.0% | source | — | — | |
| 85 | Kimi K2 0905 Estimated Moonshot AI | 26.3% | source | — | — | |
| 86 | GPT-4 Turbo OpenAI | 25.1% | source | — | — | |
| 87 | MiniMax-M2 Estimated MiniMax | 25.0% | source | — | — | |
| 88 | Claude 3 Opus Estimated Anthropic | 23.5% | source | — | — | |
| 89 | Llama 3.1-405B Meta | 23.0% | source | — | — | |
| 90 | o3-mini Estimated OpenAI | 22.8% | source | — | — | |
| 91 | Grok-2 (Dec 2024) xAI | 22.7% | source | — | — | |
| 92 | Mistral Large 2 (Nov 2024) Mistral | 22.5% | source | — | — | |
| 93 | GPT OSS 120B Supported OpenAI | 22.1% | source | — | — | |
| 94 | Mistral Large 3 Mistral | 20.4% | source | — | — | |
| 95 | Llama 3.3 70B Meta | 19.9% | source | — | — | |
| 96 | DeepSeek-V3 DeepSeek | 18.9% | source | — | — | |
| 97 | o1-mini OpenAI | 18.1% | source | — | — | |
| 98 | GPT-4o (2024-08-06) Supported OpenAI | 17.8% | source | — | — | |
| 99 | Command R+ Cohere | 17.4% | source | — | — | |
| 100 | GPT-4o mini OpenAI | 10.7% | source | — | — |