Meerada Grade — Live Board
Continuously measured on verifiable tasks · quality & cost per accepted task · every grade shows its confidence · 2026-09-09 17:14Z
| # | Model | Grade | Quality (Wilson 95%) | n | CPAT | TTAT | Confidence |
|---|
| 🥇 |
qwen/qwen3.8-27b |
83 |
83.3% [66-93] |
30 |
$0.00062 |
0.5s |
provisional |
| 🥈 |
openai/gpt-oss-120b |
78 |
77.8% [65-87] |
54 |
$0.00002 |
0.4s |
provisional |
| 🥉 |
groq/compound-mini |
66 |
65.8% [50-79] |
38 |
$0.00017 ~ |
0.9s |
provisional |
| 4 |
groq/compound |
62 |
62.0% [48-74] |
50 |
$0.00047 ~ |
2.7s |
provisional |
| 5 |
openai/gpt-oss-20b |
59 |
58.6% [47-69] |
70 |
$0.00006 |
0.8s |
provisional |
| 6 |
allam-2-7b |
34 |
33.8% [24-45] |
74 |
$0.00011 ~ |
1.3s |
provisional |