Deploy KAT-Coder-V2.5-Dev
KAT-Coder-V2.5-Dev by Kwaipilot on GPU capacity reserved for you: a private OpenAI-compatible endpoint, every request processed on European GPUs and logged in your audit trail, billed per GPU-hour on a monthly commitment.
- Maker
- Kwaipilot
- Params
- 35B
- Context
- 256K
- Architecture
- MoE · 256 experts, 8 active
- License
- Apache 2.0
- Released
- 2026-07-23
Footprint
Minimum to load the published checkpoint with a small KV cache. Long contexts and high concurrency need more; quantizing a 16-bit checkpoint needs less. You don't pick hardware on sference: you choose a performance tier and we size and run the deployment.
License
Apache 2.0Commercial use allowed
Permissive: commercial use, modification, and private deployment allowed; keep the license and notices. Read the license. Not legal advice.
FAQ
Can I run KAT-Coder-V2.5-Dev on a dedicated deployment in Europe?
Yes. sference deploys KAT-Coder-V2.5-Dev on GPU capacity reserved for you, behind a private endpoint, with every request processed on European GPUs and logged in your audit trail. Tell us your expected volume and latency targets with the form on this page and we come back with a sizing and a quote.
How many GPUs does KAT-Coder-V2.5-Dev need?
The published checkpoint is about 69 GB (BF16), so loading it takes at least 2 GPUs with 80 GB of memory each, or 1 GPU with 141 GB each. Long contexts and high concurrency need more for the KV cache; quantization needs less. You don't pick hardware on sference: you choose a performance tier and we size the deployment.
Is KAT-Coder-V2.5-Dev licensed for commercial use?
KAT-Coder-V2.5-Dev is published under the Apache 2.0 license. Permissive: commercial use, modification, and private deployment allowed; keep the license and notices.
How is a dedicated deployment priced?
Per GPU-hour on a monthly commitment, sized to your throughput and latency targets.
Related models
Request KAT-Coder-V2.5-Dev
Tell us roughly how much traffic you expect. A fine-tune of KAT-Coder-V2.5-Dev? Mention it in the notes.
- 01We size it
A performance tier, the capacity behind it, and a quote for your volume.
- 02We benchmark it
Bring one workload with a sample of real traffic; we come back with latency and throughput numbers.
- 03It goes live
We provision the deployment. You call it with the same API and model id as before.
Specs from Hugging Face, synced 2026-10-27. Machine-readable version: https://sference.com/models/Kwaipilot/KAT-Coder-V2.5-Dev.md