Atlas

Benchmarks

← All benchmarks

MMLU Pro - Computer Science

General QA · 2024-06-03

The Computer Science split of MMLU Pro. Its guessing floor averages the reciprocal option count across all 410 subject questions in official dataset revision b189ec765aa7ed75c8acfea42df31fdae71f97be. This component remains distinct from the parent aggregate.

Top models (higher is better)

ModelScore
Claude Fable 5.194.9
Opus 4.792.7
Opus 4.892.7
Gemini 3.1 Pro Preview92.7
Gemini 3.8 Flash92.7
Gemini 3.7 Flash92.0
Gemini 3.5 Flash91.7
Gemini 3.6 Flash91.7
Grok 4.691.7
Gemini 3 Pro Preview91.5
Kimi K2.691.2
Kimi K391.2
Qwen3.7-Max (source unspecified)91.2
Gemini 3 Flash Preview91.0
GPT-5.6 Terra91.0
Grok 4.591.0
Qwen3.5 Plus (2026-02-15)91.0
Qwen3.8-Max91.0
Opus 4.690.7
GPT-5.6 Sol90.7
Muse Spark 1.190.7
Muse Spark 1.290.7
DeepSeek-V4-Flash-073190.2
GLM-5.290.2
Sonnet 4.690.0
DeepSeek V4 Pro 081390.0
Muse Spark90.0
GPT-5.189.8
GPT-5.489.8
GPT-5.589.8
GPT-5.6 Luna89.8
Nemotron 3 Ultra 550B A55B (NVFP4)89.8
Qwen3.6 Plus (2026-04-02)89.8
Opus 4.189.5
Sonnet 4.589.5
GLM-5.189.5
Sonnet 589.0
DeepSeek-V4-Pro89.0
Gemini 3.1 Flash-Lite Preview89.0
GLM-5.389.0
Loading Atlas data…