- qwen2.5-1.5b-instruct6 games · 67% won · $0/run · 19 s/run · provisional1,035± 268
- llama-3.2-1b-instruct6 games · 33% won · $0/run · 18 s/run · provisional965± 268
Ratings are Bradley–Terry Elo from Bench, with a 95% interval; a model with fewer than ten games is provisional.
| Model | Quality | Games | Won | Cost/run | Speed/run |
|---|---|---|---|---|---|
| qwen2.5-1.5b-instruct | 1035 ± 268 | 6 | 67% | $0 | 19 s |
| llama-3.2-1b-instruct | 965 ± 268 | 6 | 33% | $0 | 18 s |
Latest runs0
No run of this task kind has been published yet.
Download the dataset JSONL from Bench.