| Nova Lite 1.0 | Llama 3.3 70B Instruct | |
|---|---|---|
| Input, per 1M tokens | $0.06 | $0.10 |
| Output, per 1M tokens | $0.24 | $0.32 |
| Context window | 300K | 131K |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $0.0384 | $0.0632 |
| Chat reply (2,000 in, 400 out), per 1,000 | $0.216 | $0.328 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $0.60 | $0.96 |
| Confidence signal | no | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | not measured | 305 |
Which one should carry your traffic?
If Nova Lite 1.0 answers your requests as well as Llama 3.3 70B Instruct, moving that traffic saves 34% on chat-shaped requests. Averages cannot tell you whether it does for your requests. Nova Lite 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 Lite 1.0 profile · Llama 3.3 70B Instruct profile · Measure it on your traffic