Side-by-side technical and economic comparison between OpenAI's GPT-5.6 Luna and Alibaba / Qwen's Qwen 2.5 72B Instruct — Together AI.
• Qwen 2.5 72B Instruct — Together AI: Classified as a Legacy Reference model (Deprecated on Together AI (Reference)).
| Metric | GPT-5.6 Luna | Qwen 2.5 72B Instruct — Together AI | Advantage |
|---|---|---|---|
| Provider Organization | OpenAI | Alibaba / Qwen | — |
| Standard Input / 1M | $0.20 | $0.35 | GPT-5.6 Luna (43% lower) |
| Cached Input / 1M | $0.02 | $0.175 | GPT-5.6 Luna lower |
| Output / 1M | $1.20 | $0.40 | Qwen 2.5 72B Instruct — Together AI lower |
| Context Window | 1.05M | 128k | GPT-5.6 Luna (1.05M) |
| Max Generation Tokens | 16.4k | 8.2k | GPT-5.6 Luna |
| Tokenizer Family | OpenAI o200k_base (200k vocabulary) | Qwen Byte-level BPE (~152k vocabulary) | — |
Opt for GPT-5.6 Luna when your engineering requirements prioritize OpenAI's ecosystem, specific tokenizer efficiencies (Exact BPE (o200k_base)), or when your expected prompt-to-completion ratios favor its $0.2/M input rate.
Opt for Qwen 2.5 72B Instruct — Together AI when looking for Alibaba / Qwen's tooling integration, specific context window depth (128k tokens), or when output generation volume favors its $0.4/M rate.