Summary
- Answers 48 of 80 verification questions correctly (60.0%), placing #38 of 72 models tested on the current suite.
- Generates 62 output tokens per second, faster than 28% of the models we measure.
- Starts responding after 1.1 s (median time to first token), including thinking time.
- Costs $0.0033 per question as billed, #26 cheapest of 29 models priced $0.15 to $1 per 1M tokens.