| GPT-6 Sol | gpt-oss-20b | |
|---|---|---|
| Input, per 1M tokens | $2.00 | $0.018 |
| Output, per 1M tokens | $10.00 | $0.09 |
| Context window | 1.05M | 131K |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $1.30 | $0.0117 |
| Chat reply (2,000 in, 400 out), per 1,000 | $8.00 | $0.072 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $21.00 | $0.189 |
| Confidence signal | no | yes |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | 330 | 313 |
Which one should carry your traffic?
If gpt-oss-20b answers your requests as well as GPT-6 Sol, moving that traffic saves 99% on chat-shaped requests. Averages cannot tell you whether it does for your requests. gpt-oss-20b exposes a confidence signal, so Gimbix can let it answer first and hand unsure cases to GPT-6 Sol.
GPT-6 Sol profile · gpt-oss-20b profile · Measure it on your traffic