Atlas

Benchmarks

← All benchmarks

SAGE - Linear Algebra

Math · 2025-10-08

The Linear Algebra split of SAGE. This child benchmark separates a source-reported subtask or subtrack from the parent aggregate so scores at different grains do not share one benchmark_slug.

Top models (higher is better)

ModelScore
Opus 4.878.3
Opus 4.777.0
Claude Opus 576.1
Opus 4.675.4
GPT-5.574.4
Gemini 3.1 Pro Preview74.2
Gemini 4 Argon74.0
Gemma 4 31B IT72.6
Kimi K372.2
Claude Fable 571.5
GPT-5.4 Mini71.5
Claude Fable 5.171.4
Gemini 3 Flash Preview71.0
Kimi K2.571.0
Claude Sonnet 5.570.8
Opus 4.570.6
GPT-5.6 Sol70.4
Gemini 3.6 Flash69.9
GPT-5.269.6
Qwen3.8-Max69.6
Gemini 3.7 Flash69.5
Sonnet 569.0
Muse Spark 1.168.9
GPT-5.468.8
GPT-5.168.6
GPT-6 Luna68.5
Gemini 3.5 Flash68.3
Kimi K2.668.1
Claude Opus 5.566.9
MiniMax M366.8
Qwen3.8 27B65.9
Gemini 3 Pro Preview65.1
GPT-6 Sol64.1
GPT-5.6 Terra64.0
Llama 4 Maverick Instruct63.9
Gemini 3.5 Flash-Lite63.7
GPT-5.6 Luna63.2
Muse Spark 1.263.2
Sonnet 4.663.2
MiMo-V2.6-Pro62.9
Loading Atlas data…