LLM Releases
← Catalog

Kimi-Linear-48B-A3B-Instruct

Available
Moonshot AIOpen source

MIT-licensed hybrid linear-attention model using Kimi Delta Attention, built for million-token contexts with much lower KV-cache usage.

Specifications

License
Open source · MIT
Weights
Downloadable
Architecture
hybrid
Parameters
48B · 3B active
Context window
1.0M tokens
Max output
Knowledge cutoff
Price (in / out, $/M)
Modalities
TextCode

Benchmarks

No benchmark scores recorded yet. Spotted some? Submit a correction.

Vendor-reported figures are claims until independently verified. See methodology.