MM1.0 methodology and benchmark scoring
Understand the context behind ModelMeter results, benchmark scales, and evaluation sources.
MM1.0
ModelMeter benchmark. Score scales based on vibe, accuracy, reasoning, chat, and long context. The score is calculated via weighted prompts, with each prompt containing at least one test from each category.
Scale 0–100. Higher is better.
ModelMeter benchmark methodology. The score is calculated via weighted prompts, with each prompt containing at least one test from each category. The score scales based on vibe, accuracy, reasoning, chat, and long context.
- Reasoning: 0–100
- Coding: 0–100
- Chat: 0–100
- Long Context: 0–100
- Vibe: 0–100
- Accuracy: 0–100
- General Knowledge: 0–100