KA

Umba 5.6.600_bf16

KatGGUF by TomkielUpdated 2026-10-09
Download on Hugging Face
Parameters34.7B
Context256K
Layers40
Architectureqwen35moe
Smallest quant16.4 GB
Quick answer

Umba 5.6.600_bf16 needs about 17.6 GB of memory at its main file (a 16.4 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 2 quants, from 16.4 GB to 25.1 GB. Longer context needs more: at 32K tokens the KV cache alone adds about 2.7 GB.

Umba 5.6.600: 34.7B parameter Qwen3.5 MoE model for coding and reasoning

Umba 5.6.600 is a 34.7B parameter Qwen3.5 MoE model fine‑tuned for agentic software engineering. It builds on Kwaipilot/KAT-Coder-V2.5-Dev and supports coding, reasoning, and tool use. The model is text‑only and does not include vision components.

Good at

  • Coding tasks
  • Reasoning with tools
  • Agentic software engineering

Limits

  • No vision support
  • Requires large VRAM (≥17.6 GB at 8k context)
  • Apache‑2.0 license

Best for

  • Automated code generation
  • Tool‑augmented reasoning
  • Software engineering research

Who it’s for. Developers and researchers needing a large MoE model for coding and reasoning tasks. The model is open‑source under Apache‑2.0 and is suitable for local deployment with sufficient GPU memory.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—16.4 GB 17.6 GB
Too big Get file →
—25.1 GB 26.3 GB
Too big Get file →

The 16.45 GB GGUF file needs about 17.6 GB of VRAM at 8k context, suitable for GPUs with 18 GB or more. The 25.11 GB file requires about 26.3 GB of VRAM, suitable for 32 GB GPUs. Choose the smaller file for tighter memory budgets.

What GPU do you need for Umba 5.6.600_bf16?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: the best quant that fits is the main file (16.4 GB file, about 17.6 GB in use).
  • 32 GB VRAM: the best quant that fits is the main file (25.1 GB file, about 26.3 GB in use).

Specifications

Architecture from the GGUF header

Architectureqwen35moe
Parameters34.7B (approximate, mixture of experts)
Layers40
Trained context262,144 tokens
KV cache (fp16)80 KB per token

Source

Quantized byTomkiel
FamilyKat
Licenseapache-2.0
Downloads (HF)752
Last updated2026-10-09

Frequently asked questions about Umba 5.6.600_bf16

How much VRAM does Umba 5.6.600_bf16 need?

About 17.6 GB at its main file with an 8K context window. That is the 16.4 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Umba 5.6.600_bf16 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 17.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Umba 5.6.600_bf16 GGUF?

It depends on the quant: file is 16.4 GB, file is 25.1 GB.

What context length does Umba 5.6.600_bf16 support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.7 GB at 8K and 2.7 GB at 32K.

Where can I download Umba 5.6.600_bf16 GGUF?

The files are on Hugging Face in the Tomkiel/umba-5.6.600 repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Umba 5.6.600_bf16 use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run Umba 5.6.600 on a GPU with 8 GB VRAM?

No. The model requires at least 17.6 GB of VRAM at 8k context.

Does Umba 5.6.600 support vision or multimodal inputs?

No, it is a text‑only model and does not include vision components.

Under what license is Umba 5.6.600 released?

It is released under the Apache‑2.0 license.

Other GGUFs of KAT-Coder-V2.5-Dev

Similar models

Browse all →

More Kat models

All Kat →

Other large models

All large models →