| Nova Premier 1.0 | DeepSeek V4.1 Flash | |
|---|---|---|
| Input, per 1M tokens | $2.50 | $0.0295 |
| Output, per 1M tokens | $12.50 | $0.60 |
| Context window | 1M | 1.05M |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $1.63 | $0.0237 |
| Chat reply (2,000 in, 400 out), per 1,000 | $10.00 | $0.299 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $26.25 | $0.536 |
| Confidence signal | no | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | not measured | 316 |
Which one should carry your traffic?
If DeepSeek V4.1 Flash answers your requests as well as Nova Premier 1.0, moving that traffic saves 97% on chat-shaped requests. Averages cannot tell you whether it does for your requests. DeepSeek V4.1 Flash exposes a confidence signal, so Gimbix can let it answer first and hand unsure cases to Nova Premier 1.0.
Nova Premier 1.0 profile · DeepSeek V4.1 Flash profile · Measure it on your traffic