Current recommendation
Separates pure token-dollar efficiency from premium model preference.
Source intelligence
Imported rows drive rankings; researched sources show the next best places to improve coverage and cost signal.
Benchmark coverage
Each benchmark family shows current rows, pricing coverage, source status, and whether it is active in the score.
Performance vs projected token cost
Higher and farther left means more benchmark performance per token dollar.
Compare set
Select up to four rows to compare efficiency, score, and projected workload cost.
Ranked models
Default rank is adjusted benchmark points per projected token dollar.
| Compare | Rank | Model | Composite | Evidence | Quality | Benchmarks | Published cost | Profile cost | Cost / point | Efficiency |
|---|