token.app › Compare › Qwen3.5 397B A17B vs GPT-5.6 Luna
Qwen3.5 397B A17B vs GPT-5.6 Luna
GPT-5.6 Luna is 3.9× cheaper per token on a blended 3:1 basis. GPT-5.6 Luna accepts the larger context window (1.1M).
Head to head
Blended price uses a 3:1 input:output mix
| Qwen3.5 397B A17B | GPT-5.6 Luna | |
|---|---|---|
| Provider | Alibaba | OpenAI |
| Input $/1M | $0.390 | $0.100 |
| Output $/1M | $2.34 | $0.600 |
| Blended $/1M | $0.877 | $0.225 |
| Context window | 262K | 1.1M |
| Max output | 66K | 128K |
| Released | Feb 2026 | Jul 2026 |
| Licence | Open weights | Proprietary |
Choose Qwen3.5 397B A17B if…
- Open weights — self-hostable
Choose GPT-5.6 Luna if…
- Costs less per token — $0.225 vs $0.877 blended
- Larger context window (1.1M)
- Newer model (released Jul 2026)
FAQ
Which is better, Qwen3.5 397B A17B or GPT-5.6 Luna?
We do not hold independently-run benchmark scores covering both Qwen3.5 397B A17B and GPT-5.6 Luna, so we make no quality claim. Their pricing and specifications are compared on this page.
Is Qwen3.5 397B A17B cheaper than GPT-5.6 Luna?
GPT-5.6 Luna is cheaper: $0.225 versus $0.877 per million tokens blended at 3:1 input:output. Input alone: $0.390 vs $0.100. Output alone: $2.34 vs $0.600.
What context windows do Qwen3.5 397B A17B and GPT-5.6 Luna have?
Qwen3.5 397B A17B: 262,144 tokens. GPT-5.6 Luna: 1,050,000 tokens.
Who makes Qwen3.5 397B A17B and GPT-5.6 Luna?
Qwen3.5 397B A17B is made by Alibaba. GPT-5.6 Luna is made by OpenAI.
Related comparisons