| GPT-5.4 Nano | Qwen3.8 27B | |
|---|---|---|
| Input, per 1M tokens | $0.20 | $0.42 |
| Output, per 1M tokens | $1.25 | $3.00 |
| Context window | 400K | 1M |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $0.133 | $0.282 |
| Chat reply (2,000 in, 400 out), per 1,000 | $0.90 | $2.04 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $2.23 | $4.86 |
| Confidence signal | no | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | not measured | 264 |
Which one should carry your traffic?
If GPT-5.4 Nano answers your requests as well as Qwen3.8 27B, moving that traffic saves 56% on chat-shaped requests. Averages cannot tell you whether it does for your requests. GPT-5.4 Nano does not expose a confidence signal through OpenRouter, so Gimbix would test it as a full replacement rather than a first answer.
GPT-5.4 Nano profile · Qwen3.8 27B profile · Measure it on your traffic