| Nova Micro 1.0 | Gemma 4 31B | |
|---|---|---|
| Input, per 1M tokens | $0.035 | $0.09 |
| Output, per 1M tokens | $0.14 | $0.34 |
| Context window | 128K | 262K |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $0.0224 | $0.0574 |
| Chat reply (2,000 in, 400 out), per 1,000 | $0.126 | $0.316 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $0.35 | $0.89 |
| Confidence signal | no | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | not measured | 304 |
Which one should carry your traffic?
If Nova Micro 1.0 answers your requests as well as Gemma 4 31B, moving that traffic saves 60% on chat-shaped requests. Averages cannot tell you whether it does for your requests. Nova Micro 1.0 does not expose a confidence signal through OpenRouter, so Gimbix would test it as a full replacement rather than a first answer.
Nova Micro 1.0 profile · Gemma 4 31B profile · Measure it on your traffic