LLM Model Analysis
Price versus benchmark performance, scored.
Lower = cost matters less · Higher = cost matters more
Share of input read from cache. New context uses cache-write pricing where published. See “how scores work” for assumptions.
Repeated context reads count as input; reasoning counts as output. These are workload assumptions, not measured cost per task.
best_valuebest value
Best Value
—
—
best_performancetop score
Best Performance
—
—
cheapestlowest cost
Cheapest
—
—
most_expensivehighest cost
Most Expensive
—
—
performance_leaderboard
performance_vs_cost_scatter
model_rankings_bar
model_comparison_radar
Click a model in the table, charts, or leaderboard to compare its profile against the filtered average.
Shift-click a chart point to add it to the compare set.
Tap a model in the table, charts, or leaderboard to compare its profile against the filtered average.
Use the + button on a leaderboard or table row to add models to the compare set.
model_comparison_dataset
| Provider | Model | Input $/1M | Output $/1M | Cache $/1M | Blended $/1M | LiveBench | AA Score | Performance | Value | Compare |
|---|
model_comparison
Swipe to see each selected model