Kimi K2.5 vs Gemini 3 Pro Image
Kimi K2.5 is cheaper on output tokens, while Kimi K2.5 offers a larger context window. Choose Kimi K2.5 or Gemini 3 Pro Image based on the trade-off between cost, context, and the benchmarks that matter for your use case.
| Spec | Kimi K2.5 | Gemini 3 Pro Image |
|---|---|---|
| Provider | MoonshotAI | NanoGPT |
| Input / 1M tokens | $0.50 | $2.00 |
| Output / 1M tokens | $2.50 | $12.00 |
| Context window | 262K | 66K |
| Parameters | 1T | — |
| Open weights | Yes | No |
| Released | Jan 2026 | Nov 2025 |
| LiveBench score | 69.2 | — |
| SWE-bench Multilingual score | 67.3 | — |
| SWE-bench Verified score | 70.8 | — |
| Epoch Capabilities Index score | 148 | — |
| LMArena (Chatbot Arena) Elo score | 1445 | — |
| LMArena WebDev Elo score | 1437 | — |
| LMArena Vision Elo score | 1269 | — |
FAQ
Is Kimi K2.5 or Gemini 3 Pro Image cheaper?
Kimi K2.5 is cheaper on output tokens ($2.50 vs $12.00 per 1M).
Which has the larger context window, Kimi K2.5 or Gemini 3 Pro Image?
Kimi K2.5 has the larger context window (262K tokens).
Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.