LLM
GA
Open weights
Kimi K2.5
Open-weight agentic coding and tool use
Release2026-04
Context262K
Max output—
Parameters1T
Local / VRAM900–1200 GB
LicenseOpen weights
Overview
Kimi K2.5 is a model in the LLM category from Moonshot AI (Kimi). It was released in 2026-04. The vendor is based in China. Best for: Open-weight agentic coding and tool use.
Strengths
- 1T MoE
- Modified MIT
- Visual input
Limits
- Large hardware footprint
- 262K context
Specifications
| Vendor | Moonshot AI (Kimi) |
|---|---|
| Category | LLM |
| Release | 2026-04 |
| Status | GA |
| License | Open weights |
| License name | Modified MIT |
| Commercial use | Allowed |
| Parameters | 1T |
| Active parameters | 32B |
| Architecture | MoE transformer |
| Context window | 262K |
| Input modalities | Text, Image |
| Output modalities | Text |
| Prompt types | Text |
| Negative prompts | No |
| Max outputs | 1 |
| Local deployment | Yes |
| VRAM | 900–1200 GB |
| System RAM | 1024–2048 GB |
| API access | Yes |
| Pricing | Free weights / API |
| Confidence | High |
| Vendor country | China |
| Vendor type | Open lab |
Pricing & hardware
Free weights / API
This model can be self-hosted or run locally. Approximate GPU memory: 900–1200 GB. Approximate system RAM: 1024–2048 GB. API access is available.