SShogo Bench

AlgoTune

Speed up numerical algorithms; score is speedup.

Agent Index weight
0%
shown for context only
Models tested
18
from the benchmark’s leaderboard and reports
Normalised for the index
1 → 0, 2.5 → 100
Official site

Results

#ModelScoreNormalisedHarnessCost/taskDate
1GPT-5.2 Supported
OpenAI
2.05×source——
2Gemini 3.1 Pro Preview Supported
Google
2.02×source——
3GPT-5.4 Supported
OpenAI
1.85×source——
4Gemini 3 Pro Preview Supported
Google
1.83×source——
5Claude Opus 4.5 (latest) Supported
Anthropic
1.77×source——
6o4-mini Estimated
OpenAI
1.72×source——
7DeepSeek-R1 Estimated
DeepSeek
1.70×source——
8GPT-5 Supported
OpenAI
1.67×source——
9Claude Sonnet 4.5 (latest) Supported
Anthropic
1.52×source——
10GLM-4.5
Z.ai (Zhipu)
1.52×source——
11Gemini 2.5 Pro Supported
Google
1.51×source——
12Claude Opus 4.6 Supported
Anthropic
1.47×source——
13Qwen3 Coder 480B A35B Estimated
Alibaba (Qwen)
1.44×source——
14GPT OSS 120B Supported
OpenAI
1.41×source——
15GPT-5 Mini Estimated
OpenAI
1.38×source——
16Claude Opus 4.1 (latest) Supported
Anthropic
1.34×source——
17Claude-Opus-4 Supported
Anthropic
1.33×source——
18GPT-5 Pro
OpenAI
1.31×source——