Skip to content

Benchmark

SnakeBench

Scores are percentages.

Models measured
260
Scores recorded
267
Sources
1
IQ weight
1.32

Score distribution

22.1%
Mean
22.5%
Median
7.3%
Std dev
13.2%
10th percentile
32.2%
90th percentile
1.2%
Min
1.2%38.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.32 for chat models, measured over 254 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
151GPT-5.6 Luna· SnakeBench data20.7%
152Magistral Mediumthinking· SnakeBench data20.7%
153Llama 4 Scout· SnakeBench data20.7%
154DeepSeek V3· SnakeBench data20.6%
155Hermes 4 405B· SnakeBench data20.5%
156Qwen: Qwen3 VL 8B Thinking· SnakeBench data20.5%
157Qwen Plus· SnakeBench data20.4%
158MiMo V2 Omni· SnakeBench data20.4%
159Qwen 3 235B A22B· SnakeBench data20.4%
160Hermes 4 70B· SnakeBench data20.4%
161Perceptron Mk1· SnakeBench data20.3%
162Qwen Max· SnakeBench data20.3%
163Llama 3 70B Instruct· SnakeBench data20.3%
164Phi 4 Multimodal Instruct· SnakeBench data20.3%
165Gemini Flash Latest· SnakeBench data20.2%
166LongCat Flash Chat· SnakeBench data19.9%
167DeepSeek V3.2· SnakeBench data19.9%
168Qwen2.5 VL 72B Instruct· SnakeBench data19.8%
169Qwen3 14B· SnakeBench data19.7%
170Qwen 2.5 Coder 32B Instruct· SnakeBench data19.7%
171MiniMax M2.1· SnakeBench data19.7%
172GPT-4o mini· SnakeBench data19.6%
173Trinity Large Thinking· SnakeBench data19.6%
174InternVL3 78B· SnakeBench data19.4%
175Olmo 3 32B Think· SnakeBench data19.3%
176gemma-2-9b· SnakeBench data19.2%
177Gemma 3 12B IT· SnakeBench data19.2%
178Magistral Small 25092506· SnakeBench data19.1%
179Sonar Pro· SnakeBench data19%
180Ministral 8B· SnakeBench data18.7%
181LlamaGuard 2 8B· SnakeBench data18.6%
182Qwen3.6 27B· SnakeBench data18.4%
183ERNIE 4.5 21B A3Bthinking· SnakeBench data18.4%
184INTELLECT-3· SnakeBench data18.3%
185Llama 3.1 8B Instruct· SnakeBench data18.2%
186Qwen3 Coder Flash· SnakeBench data18.2%
187Llama Guard 4 12B· SnakeBench data18.2%
188OLMo 2 32B Instruct· SnakeBench data18.1%
189Nemotron Nano 9b V2· SnakeBench data18.1%
190Llama 3.2 11B Vision Instruct· SnakeBench data18%
191KAT-Coder-Pro V1· SnakeBench data18%
192Voxtral Small 24B2507· SnakeBench data18%
193GLM 4.7 Flash· SnakeBench data17.9%
194Gemma 4 31B IT· SnakeBench data17.9%
195Command A· SnakeBench data17.9%
196Phi 4· SnakeBench data17.8%
197Jamba Large 1.7· SnakeBench data17.8%
198Qwen2.5 72B Instruct· SnakeBench data17.8%
199Qwen2.5 Coder 7B Instruct· SnakeBench data17.5%
200Pixtral 12B· SnakeBench data17%

Showing 151–200 of 267 results.

Data credits: SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence)