Dial in difficulty, context size and iteration budget — get a ranked shortlist with quotas, strengths and community signal.
Hardest job difficulty the model handles well, based on the Artificial Analysis Intelligence Index. 5 = frontier reasoning; 1 = light tasks.
Context capacity and long-context retention. 1 = Small, 2 = Average, 3 = Extra (long documents and large codebases).
How efficiently the model completes a task with fewer retries and hallucinations. 1 = very inefficient; 4 = very efficient.
Req / 5h and Req / mo show the plan limits. Unlimited means no stated cap.
Estimated sustainable daily requests: MIN(req/5h × 24 / 5, req/mo ÷ 30). The lower limit wins.
Rank is the current sort position. Difficulty, size, and effectiveness are minimums; daily requests is the minimum capacity needed.