Summary
- Answers 58 of 80 verification questions correctly (72.5%), placing #21 of 72 models tested on the current suite.
- Generates 137 output tokens per second, faster than 75% of the models we measure.
- Starts responding after 1.2 s (median time to first token), including thinking time.
- Costs $0.00025 per question as billed, #3 cheapest of 29 models priced $0.15 to $1 per 1M tokens.