MiniMax M2.5 vs DeepSeek V4.1 Flash 0910 Uncensored Thinking
MiniMax M2.5 is cheaper on output tokens, while DeepSeek V4.1 Flash 0910 Uncensored Thinking offers a larger context window. Choose MiniMax M2.5 or DeepSeek V4.1 Flash 0910 Uncensored Thinking based on the trade-off between cost, context, and the benchmarks that matter for your use case.
| Spec | MiniMax M2.5 | DeepSeek V4.1 Flash 0910 Uncensored Thinking |
|---|---|---|
| Provider | MiniMax | NanoGPT |
| Input / 1M tokens | $0.30 | $0.33 |
| Output / 1M tokens | $1.20 | $1.32 |
| Context window | 205K | 524K |
| Parameters | — | — |
| Open weights | Yes | Yes |
| Released | Feb 2026 | Oct 2026 |
| LiveBench score | 60.3 | — |
| SWE-bench Verified score | 75.8 | — |
| LMArena (Chatbot Arena) Elo score | 1359 | — |
| LMArena WebDev Elo score | 1386 | — |
| Epoch Capabilities Index score | 147 | — |
FAQ
Is MiniMax M2.5 or DeepSeek V4.1 Flash 0910 Uncensored Thinking cheaper?
MiniMax M2.5 is cheaper on output tokens ($1.20 vs $1.32 per 1M).
Which has the larger context window, MiniMax M2.5 or DeepSeek V4.1 Flash 0910 Uncensored Thinking?
DeepSeek V4.1 Flash 0910 Uncensored Thinking has the larger context window (524K tokens).
Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.