Reasoning
Multi-step problems where the answer depends on working through intermediate steps rather than recalling a fact.
O router mapeia uma tarefa para modelos candidatos e depois aplica a sua política: primeiro a região, depois o preço, depois o modelo com o melhor resultado documentado.
Prefer models that expose a reasoning mode and publish AIME or GPQA scores, and budget for the longer output those modes produce.
{"model":"llmeu-auto","messages":[{"role":"user","content":"…"}],"llmeu":{"task":"reasoning"}}Recomendado
No first-party endpoint yet
No first-party endpoint serves this task yet. The models listed below are catalogued for reference and run under their own provider’s terms.