Kimi K3 Hits Together AI: 3T Parameter MoE with 1M Token Context
zainhas · x · 2026-07-28
Moonshot AI's latest model, Kimi K3, is now available on the Poe platform via Together AI.
Based on platform details, Kimi K3 introduces the following key features:
- Scale & MoE: Boasts 2.8 trillion total parameters (3T class) with a Mixture-of-Experts (MoE) architecture, activating 16 out of 896 experts per token.
- Architectural Changes: Incorporates Kimi Delta Attention and Attention Residuals, modifying information flow across sequence length and model depth.
- Native Capabilities: Features native vision (capable of reading screenshots, charts, and documents) and a massive 1-million-token context window.
- Use Case: Designed for long-horizon tasks such as extended engineering sessions across large repositories and multi-step research. It launches with maximum thinking effort enabled and open-licensed weights.
More from Models
- A blunt model note says Sol complicates everything and Opus 5 breaks things — JsonBasedman · 2026-07-28
- Fireworks says Kimi K3 matches Opus 5 closely on 663 coding tasks while costing 2.3x less — lqiao · 2026-07-28
- Anthropic rumors point to a larger internal teacher model and a near-K3 public stack — teortaxesTex · 2026-07-28
- Kimi report reveals a wide internal benchmark suite for coding and agent skills — stochasticchasm · 2026-07-28
- Claude is still being called the most steerable model set, despite its weirdness — sloppenheimer · 2026-07-28
- Frontend Code Arena: Opus 5 Max Takes #1, Kimi K3 Max Follows Closely — arena · 2026-07-28