List your company

Code Generation on MBPP (MA, MI, Error Rate)

0.68MAPROBELLM
0.25360.36430.4750.5857Feb 13, 2026

Evaluation results

MethodLinks
PROBELLMTarget Model=GPT-oss-20b2026.020.680.320.32
PROBELLMTarget Model=Deepseek-v3.22026.020.650.350.46
PROBELLMTarget Model=olmo-3-7b-ins.2026.020.590.410.76
PROBELLMTarget Model=Gemini-2.5-flash2026.020.530.470.54
PROBELLMTarget Model=granite-4.02026.020.50.50.83
PROBELLMTarget Model=Claude-3.5-sonnet2026.020.420.580.62
PROBELLMTarget Model=ministral-14b2026.020.420.580.64
PROBELLMTarget Model=Llama-3.1-8b-ins.2026.020.40.60.9
PROBELLMTarget Model=phi-42026.020.390.610.78
PROBELLMTarget Model=Grok-4.1-fast2026.020.350.650.39
PROBELLMTarget Model=GPT-4o-mini2026.020.310.690.76
PROBELLMTarget Model=devstral2026.020.270.730.5