allbenched allbenched Allbenched allbenched

Benchmarks / Coding

GSO

Making real software measurably faster: performance optimization tasks.

CategoryCoding
MetricAccuracy
SourceEpoch AI (evaluator)
LicenseCC BY 4.0
Benchmark pagegso-bench.github.io
StatusScored
Weight0.9 within Coding
FitDiscrimination 1.18, noise 0.39

Results

#ModelProviderResultDate
1 Claude Fable 5.1 Anthropic 88.2% 2026-11-11
2 GPT-6 Astra OpenAI 79.4% 2026-11-11
3 Claude Fable 5 Anthropic 78.4% 2026-11-11
4 GPT-5.6 Sol OpenAI 76.5% 2026-11-11
5 Claude Opus 4.8 Anthropic 47.1% 2026-11-11
6 Claude Opus 4.7 Anthropic 44.1% 2026-11-11
7 Claude Opus 4.7 High Anthropic 44.1% 2026-11-11
8 Claude Opus 4.6 High Anthropic 41.2% 2026-11-11
9 GPT-5.5 Extra High OpenAI 40.2% 2026-11-11
10 Claude Sonnet 5 Anthropic 37.3% 2026-11-11
11 Claude Opus 4.6 Anthropic 33.3% 2026-11-11
12 GPT-5.4 Extra High OpenAI 31.4% 2026-11-11
13 GPT-5.2 High OpenAI 27.5% 2026-11-11
14 Claude Opus 4.5 Anthropic 26.5% 2026-11-11
15 GPT-5.4 High OpenAI 25.5% 2026-11-11
16 Gemini 3.1 Pro Preview Google 22.6% 2026-11-11
17 Gemini 3 Pro Preview Google 18.6% 2026-11-11
18 Claude Sonnet 4.5 Anthropic 14.7% 2026-11-11
19 GPT-5.1 OpenAI 13.7% 2026-11-11
20 GPT-5.1 High OpenAI 13.7% 2026-11-11
21 Gemini 3 Flash Preview Google 9.8% 2026-11-11
22 o3 High OpenAI 8.8% 2026-11-11
23 Claude Opus 4 Anthropic 6.9% 2026-11-11
24 GPT-5 High OpenAI 6.9% 2026-11-11
25 Kimi K2 Instruct Moonshot AI 4.9% 2026-11-11
26 Claude Sonnet 4 Anthropic 4.9% 2026-11-11
27 Qwen3-Coder 480B-A35B Instruct Alibaba 4.9% 2026-11-11
28 Claude Sonnet 3.5 v2 Anthropic 4.6% 2026-11-11
29 Gemini 2.5 Pro Google 3.9% 2026-11-11
30 Gemini 2.5 Pro Preview 06-05 Google 3.9% 2026-11-11
31 Claude Sonnet 3.7 Anthropic 3.8% 2026-11-11
32 o4-mini High OpenAI 3.6% 2026-11-11
33 GLM-4.5-Air Zhipu AI 2.9% 2026-11-11
34 o3-mini Low OpenAI 1.3% 2026-11-11
35 o3-mini High OpenAI 1.3% 2026-11-11
36 GPT-4o (2024-11-20) OpenAI 0.0% 2026-11-11