Explorer
LLM cost comparison, ranked per task dollar
Compare LLM costs per task and rank reasoning variants by capability per measured or estimated task dollar. The default view keeps Artificial Analysis Intelligence at 40 or more. The EQ (Effiq) Score is not fixed — it changes with your requirements: profile, weights, floors, and filters all move the ranking.
Top picks right now
Loading the live matrix… Source status.
Broad capability with balanced cost. Workload: Standard mixed task.
Lock up to 4 sliders. Locked sliders stay fixed while the other sliders share the remainder. Click a preset to unlock all.
0/4 locked
| Compare and pin row controls | # | Model variant | Effiq Score | Domain eff. | Domain | Intel | Coding | Agentic | CursorBench | Task $ | Tokens/task | Task time | TPS | Latency | In $/1M | Out $/1M | Cache $/1M | Cap/$ |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Loading Effiq rankings… | ||||||||||||||||||
The default view keeps Artificial Analysis Intelligence at 40 or higher. The table ranks models by the Effiq Score. The EQ (Effiq) Score is not fixed — it changes with your requirements: the usage profile, metric weights, floors, and filters all move the ranking. The score compares capability against task cost, latency, and throughput. Approximated model variants have clear labels. Primary data sources are Artificial Analysis, OpenRouter, Cursor, and OpenCode Go.