| Kimi K2.6 | Qwen3.8 27B | |
|---|---|---|
| Input, per 1M tokens | $0.65 | $0.42 |
| Output, per 1M tokens | $3.41 | $3.00 |
| Context window | 262K | 1M |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $0.424 | $0.282 |
| Chat reply (2,000 in, 400 out), per 1,000 | $2.66 | $2.04 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $6.91 | $4.86 |
| Confidence signal | yes | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | 323 | 264 |
Which one should carry your traffic?
If Qwen3.8 27B answers your requests as well as Kimi K2.6, moving that traffic saves 23% on chat-shaped requests. Averages cannot tell you whether it does for your requests. Qwen3.8 27B exposes a confidence signal, so Gimbix can let it answer first and hand unsure cases to Kimi K2.6.
Kimi K2.6 profile · Qwen3.8 27B profile · Measure it on your traffic