Qwen3.8 27BvsGPT-6 Luna
Shareable link:
Before you run
Highlighted values are the best among your picks.
Qwen3.8 27BAlibaba· open weights Qwen vision-language model for visual reasoning, documents, and agent tasks | GPT-6 LunaOpenAI OpenAI's most efficient model for focused, high-volume tasks | |
|---|---|---|
| Intelligence | ||
| Intelligence index | 33.7 | 38.1 |
| Price | ||
| Input, $ per 1M | $0.50 | $0.10 |
| Output, $ per 1M | $3 | $0.50 |
| Typical message (500 in / 250 out) | $0.0010 | $0.00017 |
| Repeated input, cached | $0.10 · 5× cheaper | $0.01 · 10× cheaper |
| Prompts over 200K tokens | Same price | $0.20 / $0.75 |
| Limits | ||
| Context window | 1M | 1.05M |
| Longest answer | 131K tokens | 128K tokens |
| Knowledge cutoff | Not published | May 2026 |
| What it can take and do | ||
| Accepts | TextImagePDF | TextImagePDF |
| Reasoning effort | Off · Low · Med · X-high | Off · Low · Med · High · X-high · Max |
| Tool use | Yes | Yes |
| Guaranteed JSON output | Yes | Yes |
| Availability | ||
| Released | 14 Aug 2026 | 22 Sept 2026 |
| Open weights | Yes, can self-host | No |
| API providers offering it | 47 | 36 |