Skip to content
coding benchmark · included in ECI

FrontierCode

FrontierCode benchmark (score column: Main score).
Results
12
Random baseline
0.0%
Score ceiling
100%
Released
8 Jun 2026

Score by model release date

Best score by organisation

#ModelScoreRelativeEvidence
1Kimi K3 (unknown) Open source
Moonshot
44.2%Source ↗
2GLM 5.3 Open source
Z.ai
40.1%Source ↗
3GLM 5.3 Flash Open source
Z.ai
31.8%Source ↗
4Kimi K2.7 Code Open source
MoonshotAI
30.1%Source ↗
5DeepSeek V4 Pro 0813 Open source
DeepSeek
28.5%Source ↗
6GLM 5.2 Open source
Z.ai
24.5%Source ↗
7DeepSeek V4 Flash 0731 (unknown) Open source
DeepSeek
18.8%Source ↗
8DeepSeek v4 Pro (unknown thinking) Open source
DeepSeek
17.6%Source ↗
9MiniMax M3 Open source
MiniMax
14.7%Source ↗
10Inkling Open source
Thinking Machines
14.0%Source ↗
11Nemotron 3 Ultra Open source
NVIDIA
13.6%Source ↗
12Mistral Medium 3.5 Open source
Mistral
8.0%Source ↗

Caveats: settings (reasoning effort, agent scaffold) differ between rows and materially affect scores; the best reported setting per model is shown. Source: Epoch AI, CC BY 4.0.