List your company

Language Understanding on SuperBLUE

0.61MA.PROBELLM
0.32920.40210.4750.5479Feb 13, 2026

Evaluation results

MethodLinks
PROBELLMTarget Model=granite-4.02026.020.610.390.75
PROBELLMTarget Model=devstral2026.020.590.410.44
PROBELLMTarget Model=phi-42026.020.580.420.59
PROBELLMTarget Model=olmo-3-7b-ins.2026.020.580.420.69
PROBELLMTarget Model=Claude-3.5-sonnet2026.020.530.470.5
PROBELLMTarget Model=GPT-4o-mini2026.020.530.470.65
PROBELLMTarget Model=Grok-4.1-fast2026.020.50.50.39
PROBELLMTarget Model=ministral-14b2026.020.480.520.65
PROBELLMTarget Model=GPT-oss-20b2026.020.480.520.31
PROBELLMTarget Model=Llama-3.1-8b-ins.2026.020.460.540.71
PROBELLMTarget Model=Gemini-2.5-flash2026.020.450.550.46
PROBELLMTarget Model=Deepseek-v3.22026.020.340.660.43