MiniMax M3 vs LongCat 2.5 Preview
MiniMax M3 is cheaper on output tokens, while MiniMax M3 offers a larger context window. Choose MiniMax M3 or LongCat 2.5 Preview based on the trade-off between cost, context, and the benchmarks that matter for your use case.
| Spec | MiniMax M3 | LongCat 2.5 Preview |
|---|---|---|
| Provider | MiniMax | NanoGPT |
| Input / 1M tokens | $0.30 | $0.75 |
| Output / 1M tokens | $1.20 | $2.95 |
| Context window | 1M | 1M |
| Parameters | — | — |
| Open weights | Yes | No |
| Released | Jun 2026 | Sep 2026 |
| LiveBench score | 67.3 | — |
| LMArena (Chatbot Arena) Elo score | 1433 | — |
| LMArena WebDev Elo score | 1482 | — |
| LMArena Vision Elo score | 1251 | — |
| GPQA Diamond score | 90.9 | — |
| Epoch Capabilities Index score | 147 | — |
FAQ
Is MiniMax M3 or LongCat 2.5 Preview cheaper?
MiniMax M3 is cheaper on output tokens ($1.20 vs $2.95 per 1M).
Which has the larger context window, MiniMax M3 or LongCat 2.5 Preview?
MiniMax M3 has the larger context window (1M tokens).
Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.