GPT-5.6 Luna vs Qwen3 VL 235B A22B Instruct
Price and specs side by side · USD per 1M tokens · verified 2026-11-22
// side by side
| metric | GPT-5.6 Luna | Qwen3 VL 235B A22B Instruct |
|---|
| input / 1M | $0.20 | $0.20 |
| output / 1M | $1.20 | $0.88 |
| blended (3:1) | $0.45 | $0.37 |
| cache read / 1M | $0.02 | $0.11 |
| context | 1.05M | 262.1K |
| max output | 128K | 32.8K |
| input types | text · image · pdf | text · image |
| reasoning | yes | no |
| tool calling | yes | yes |
| structured output | yes | yes |
| open weights | no | yes |
| released | 2026-07-09 | 2025-09-23 |
| knowledge cutoff | 2026-02-16 | 2025-03-31 |
| lab | OpenAI | Alibaba (Qwen) |
// monthly cost scenarios
| workload | tokens in / out | GPT-5.6 Luna | Qwen3 VL 235B A22B Instruct |
|---|
| Chatbot | 10M / 2M | $4.40 | $3.76 |
| RAG app | 50M / 5M | $16.00 | $14.40 |
| Coding agent | 20M / 10M | $16.00 | $12.80 |
// verdict
- Qwen3 VL 235B A22B Instruct is 17.8% cheaper than GPT-5.6 Luna on a blended 3:1 input/output basis.
- GPT-5.6 Luna has the larger context window (1.05M vs 262.1K tokens).
- Only GPT-5.6 Luna supports built-in reasoning.
- Chatbot (10M input / 2M output tokens per month): Qwen3 VL 235B A22B Instruct costs $3.76 vs $4.40 for GPT-5.6 Luna, saving $0.64 per month.
- RAG app (50M input / 5M output tokens per month): Qwen3 VL 235B A22B Instruct costs $14.40 vs $16.00 for GPT-5.6 Luna, saving $1.60 per month.
- Coding agent (20M input / 10M output tokens per month): Qwen3 VL 235B A22B Instruct costs $12.80 vs $16.00 for GPT-5.6 Luna, saving $3.20 per month.
// GPT-5.6 Luna or Qwen3 VL 235B A22B Instruct: which one to choose
- Choose Qwen3 VL 235B A22B Instruct for cost-sensitive, high-volume workloads.
- Choose GPT-5.6 Luna for long documents and large codebases (1.05M-token context).
- Choose GPT-5.6 Luna if you need built-in reasoning.
- Choose Qwen3 VL 235B A22B Instruct if you want open weights you can self-host.