| Qwen3.8 27B | Grok 4.7 | |
|---|---|---|
| Input, per 1M tokens | $0.42 | $1.60 |
| Output, per 1M tokens | $3.00 | $4.80 |
| Context window | 1M | 500K |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $0.282 | $1.01 |
| Chat reply (2,000 in, 400 out), per 1,000 | $2.04 | $5.12 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $4.86 | $15.20 |
| Confidence signal | yes | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | 264 | not measured |
Which one should carry your traffic?
If Qwen3.8 27B answers your requests as well as Grok 4.7, moving that traffic saves 60% 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 Grok 4.7.
Qwen3.8 27B profile · Grok 4.7 profile · Measure it on your traffic