| Claude Opus 5 | Qwen3.8 27B | |
|---|---|---|
| Input, per 1M tokens | $5.00 | $0.42 |
| Output, per 1M tokens | $25.00 | $3.00 |
| Context window | 1M | 1M |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $3.25 | $0.282 |
| Chat reply (2,000 in, 400 out), per 1,000 | $20.00 | $2.04 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $52.50 | $4.86 |
| Confidence signal | no | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | not measured | 264 |
Which one should carry your traffic?
If Qwen3.8 27B answers your requests as well as Claude Opus 5, moving that traffic saves 90% 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 Claude Opus 5.
Claude Opus 5 profile · Qwen3.8 27B profile · Measure it on your traffic