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
201DeepSeek R1 0528 Qwen3 8B· SnakeBench data16.8%
202Qwen 2.5 VL 32B Instruct· SnakeBench data16.7%
203GLM 4.5· SnakeBench data16.7%
204GPT 5.4 Image 2· SnakeBench data16.6%
205Qwen3.6 Flash· SnakeBench data16.6%
206Qwen 3 235B A22B Thinking 2507· SnakeBench data16.6%
207Virtuoso Large· SnakeBench data16.5%
208Qwen3 4B· SnakeBench data16.4%
209Seed 1.6 Flash· SnakeBench data16.2%
210mai-ds-r1· SnakeBench data16.2%
211GPT-4.1 nano· SnakeBench data16%
212Qwen3 30B A3B· SnakeBench data16%
213WizardLM 2 8x22B· SnakeBench data16%
214Qwen2.5 7B Instruct· SnakeBench data16%
215Relace Apply 3· SnakeBench data15.9%
216Ring 1t· SnakeBench data15.6%
217MiniMax-M2 Her· SnakeBench data15.5%
218Maestro Reasoning· SnakeBench data15.3%
219Hunyuan A13B Instruct· SnakeBench data15.2%
220LFM2-8B-A1B· SnakeBench data15.1%
221Ling 1t· SnakeBench data15%
222Skyfall 36B V2· SnakeBench data14.9%
223Kimi Dev 72B· SnakeBench data14.9%
224GPT-5.4 mini· SnakeBench data14.9%
225Mistral Nemo· SnakeBench data14.8%
226codellama-7b-instruct-solidity· SnakeBench data14.7%
227Kimi Linear 48B A3B Instruct· SnakeBench data14.7%
228Mistral (7B) Instruct v0.3· SnakeBench data14.7%
229Mistral 7B Instruct v0.2· SnakeBench data14.6%
230Llama 3.2 3B· SnakeBench data14.2%
231ERNIE 4.5 VL 28B A3B· SnakeBench data13.9%
232jamba-mini-1.7· SnakeBench data13.8%
233Llama Guard 3 8B· SnakeBench data13.6%
234Command R7b 12 2024· SnakeBench data13.5%
235Mixtral 8x7B Instruct· SnakeBench data13.4%
236Nova Lite 1.0· SnakeBench data13.4%
237GPT Audio· SnakeBench data13.4%
238Qwen3 VL 8B Instruct· SnakeBench data13.3%
239GPT 5.4 Nano· SnakeBench data13.2%
240MiniMax 01· SnakeBench data13.2%
241Morph V3 Large· SnakeBench data13.1%
242GLM 4.1V 9B Thinking· SnakeBench data13.1%
243Phi-3 Mini 128K Instruct· SnakeBench data12.1%
244GPT-4o-mini Search Preview· SnakeBench data12%
245Granite 4.1 8B· SnakeBench data11.8%
246Hermes 2 Pro Llama 3 8B· SnakeBench data11.5%
247Reka Flash 3· SnakeBench data11.4%
248Llama 3 8B Instruct· SnakeBench data11.2%
249Olmo 3 7B Instruct· SnakeBench data10.1%
250ERNIE 4.5 21B A3B· SnakeBench data10%

Showing 201–250 of 267 results.

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