| Llama 4 Maverick | GPT-5.6 Luna | |
|---|---|---|
| Input, per 1M tokens | $0.188 | $0.20 |
| Output, per 1M tokens | $0.652 | $1.20 |
| Context window | 1.05M | 1.05M |
| Classification or routing (600 tokens in, 10 out), per 1,000 | $0.119 | $0.132 |
| Chat reply (2,000 in, 400 out), per 1,000 | $0.636 | $0.88 |
| Retrieval-augmented answer (8,000 in, 500 out), per 1,000 | $1.83 | $2.20 |
| Confidence signal | yes | no |
| Tool calling | yes | yes |
| BANKING77, correct of 400 | 294 | not measured |
Which one should carry your traffic?
If Llama 4 Maverick answers your requests as well as GPT-5.6 Luna, moving that traffic saves 28% on chat-shaped requests. Averages cannot tell you whether it does for your requests. Llama 4 Maverick exposes a confidence signal, so Gimbix can let it answer first and hand unsure cases to GPT-5.6 Luna.
Llama 4 Maverick profile · GPT-5.6 Luna profile · Measure it on your traffic