LLM rankings — OpenRouter × Ollama Cloud

Token-usage leaderboard from openrouter.ai/rankings, joined against the models your Ollama Cloud account can actually call. Set MAF_USE_RANKINGS=1 to make the smart picker prefer the highest-ranked available model per task profile.

Picker preview

Effective column shows what the picker returns for that profile right now. When MAF_USE_RANKINGS=1 is set, "effective" matches "rankings pick" (chosen via the benchmark dimension shown, or popularity when no dimension applies). Otherwise it matches "static map".

Leaderboard

Tokens column = real usage from OpenRouter. Intelligence/Coding/Agentic = Artificial-Analysis benchmark scores (higher is better; 0 = not benchmarked).