Best efficiency
—
—
Lowest task cost
—
—
Highest intelligence
—
—
Fastest output
—
—
Profile
Broad capability with balanced cost. Workload: Standard mixed task.
intelligence35%
coding15%
agentic10%
task cost30%
latency5%
throughput5%
Not enough points for Capability vs task cost.
Not enough points for Throughput vs latency.
| # | Model variant | Eff. score | Domain eff. | Domain | Intel | Coding | Agentic | Task $ | TPS | Latency | Cap/$ |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Loading efficiency matrix… | |||||||||||
Default view keeps Artificial Analysis Intelligence ≥ 40 and ranks by a transparent Efficiency Score (capability vs measured/estimated task cost, latency, and throughput). Approximated reasoning variants are labeled. Primary sources: Artificial Analysis, OpenRouter, Cursor. Usage profiles change domain evidence, weights, and workload cost assumptions.