models
Every model tested by the Omega harness, ranked by pass rate across all suites.
nvidia/nemotron-3-ultra-550b-a55b:free
openrouter
100%
1 tasks
48.9s
45.1k tokens
kimi-k3
kimi
2 tasks
2m 47s
189.8k tokens
deepseek-v4-pro+qwen3.8-max-preview
consensus
10 tasks
3m 29s
2.3M tokens
agy+MiniMax-M3+deepseek-v4-pro
36 tasks
3m 4s
7.9M tokens
default
internal
82%
49 tasks
1m 37s
2.2M tokens
qwen3.8-max-preview
qwen
70%
20 tasks
2m 55s
agy
external
64%
47 tasks
2m 24s
- tokens
deepseek-v4-pro
deepseek
39%
59 tasks
2m 45s
6.4M tokens
MiniMax-M3
minimax
31%
48 tasks
2m 34s
5.0M tokens
moonshot-v1-128k
7%
43 tasks
25m 1s
31.6M tokens
0%
4.2s