gemma4-26b
Default static recommendation from the current snapshot.
- Accept rate
- 78%
- $ / attempt
- $0.0013
- Hard fail
- 2%
- Latency
- 91.0 s
← ChoreCode leaderboard · 20260706T163525Z
Recommendation explorer
Tune quality, cost, and hard-fail tolerance locally in the browser. The data comes from static benchmark JSON; no server runtime is required.
Optimal model
default: cheapest model within 95% of best accept rate| Model | Accept rate | $ / attempt | Hard fail | Tokens / accepted | Mean latency |
|---|---|---|---|---|---|
| deepseek-v4-flash | 43% | $0.000088 | 0% | 1384 | 2.5 s |
| deepseek-v4-pro | 41% | $0.0007 | 4% | 2156 | 5.3 s |
| gemma4-26b | 78% | $0.0013 | 2% | 5411 | 91.0 s |
| glm-5.2 | 38% | $0.0009 | 0% | 1802 | 5.9 s |
| minimax-m3 | 41% | $0.0015 | 0% | 3858 | 11.8 s |
| minimax-m2.7 | 17% | $0.0008 | 3% | 6925 | 19.9 s |
| qwen3.6-27b | 48% | $0.0060 | 1% | 5646 | 157.5 s |
| kimi-k2.6 | 62% | $0.0147 | 1% | 7224 | 39.3 s |