Which agent to start, from three live sources.
| Role | Model | Credits / task | Base tasks / day | Base credits / day | Or spare buys / day |
|---|
The cost frontier from the map, rung by rung: what each next point of Intelligence costs. Click a rung to open it on the map.
The same model under the same name can simply get worse. The sparkline shows individual benchmark runs from the last 7 days — thin line raw, thick line smoothed. Single runs swing hard, so Δ7d compares the average of the first fifth of the window with the average of the last fifth. Effort is not published here — these runs use each provider's default setting.
| Model | Score | 7 days | Δ7d | min–max | Trend |
|---|
The same model is a different tool at a different effort: Opus 5.5 finishes a task in about a minute at low and seven and a half at extra high. You iterate with the worker, so the length of one loop is the length of the job. Filter by the minutes per task you will accept, or sort any column.
| Model · effort | Intelligence | Agentic | Credits / task | Min / task | First answer | Types | Drift | Copilot $ / 1M |
|---|
Cyan — the numbers line up. Take it.
Violet — a gap. You are paying a lot for a little.
Effort — always named on the model, never implied.