Grawnch/Kimi-K3.1-Pro-Coder
archived · low-signalmodel
Grawnch/Kimi-K3.1-Pro-Coder is a full 96-shard (774.97 GB / 721.75 GiB) Strata-Coder 3-Tier Mixed-Bit MoE quantization of moonshotai/Kimi-K3.1-Pro (2.8T parameters, 93 layers, 896 routed experts per MoE layer, topk=16 LatentMoE with SiTU-GLU, KDA linear attention + gated NoPE MLA full attention… Mixture-of-experts, 1270B total (896 experts, 16 per token). Context: 1M tokens (config). License: apache-2.0.
- Org
- Grawnch
- Released
- 2026-11-08
- Status
- archived · low-signal
- License
- apache-2.0 (permissive)
- Parameters
- 1270.0B
- Architecture
- MoE · 896 experts, 16/token · KimiK31ForConditionalGeneration
- Context
- 1,048,576 tokens (config.json)
- Task
- image-text-to-text · vision
- Signal
- 5 HF likes · 658 downloads
- Archived
- 2026-11-12 (no longer listed by the collector; stats as last seen)
Runs on
estimate from params; need = weights + FP16 KV cache for 8K tokens + 1GB. A sizing calculation, not tested.
- Q4 ~763.9 GB (estimated size): no listed device
- Q8 ~1351.2 GB (estimated size): no listed device
- FP16 ~2541.8 GB (estimated size): no listed device
Links
First published here 2026-11-08 · updated 2026-11-12 · markdown · JSON