Short answer: Qwen 3.8 Flash is 3.7x cheaper (0.034¢ vs 0.13¢ a chat). Qwen 3.8 Flash ranks higher overall (#20 vs #23). Qwen 3.8 Flash wins at maths, writing. MiniMax M3 wins at coding, following instructions. Qwen 3.8 Flash is the better all-rounder for the money, but MiniMax M3 is worth it if you mostly need coding, following instructions.


| Qwen 3.8 Flash | MiniMax M3 | |
|---|---|---|
| Cost per chat (NZD) | 0.034¢ | 0.13¢ |
| Chats per NZ$1 | 2,913 | 792 |
| Overall rank | #20 | #23 |
| Vendor | Alibaba | MiniMax |
| Speciality | General | Agentic |
| Reasoning | 56 | 56 |
| Code | 50 | 58 |
| Math | 52 | 36 |
| Writing | 54 | 51 |
| Instruction | 54 | 66 |
Scores are 0 to 100 percentiles blending LMArena, Epoch AI, Artificial Analysis and CrapGPT's own tests. Updated 2026-10-11.
Qwen 3.8 Flash costs 0.034¢ per typical chat versus 0.13¢ for MiniMax M3 on CrapGPT (NZD, measured token use).
Qwen 3.8 Flash ranks higher overall on CrapGPT's blended benchmark score. By category: Qwen 3.8 Flash leads at maths, writing; MiniMax M3 leads at coding, following instructions.
Yes. CrapGPT's card battle sends one prompt to two models and lets you pick the winner, paying only for the two replies.
Battle them yourself Qwen 3.8 Flash details · MiniMax M3 details