KO

Kolibri-1

KolibriGGUF by aparuselUpdated 2026-10-06
Download on Hugging Face
Parameters78.1B
Context256K
Layers50
Architecturekolibri1
Smallest quant24.5 GB
Quick answer

Kolibri-1 needs about 47.0 GB of memory at its main file (a 45.7 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 3 quants, from 24.5 GB to 83.1 GB. Longer context needs more: at 32K tokens the KV cache alone adds about 3.4 GB.

Kolibri-1: 78.1B‑parameter German/English MoE reasoning model

Kolibri-1 is a 78.1‑billion‑parameter mixture‑of‑experts model focused on German and English. It supports explicit reasoning and tool calling and is optimized for long‑context inference. The model is available in GGUF format for Metal and CPU inference.

Good at

  • Long‑context reasoning
  • Tool‑calling
  • Multi‑step problem solving

Limits

  • No vision support
  • CUDA/ROCm not supported
  • Requires large VRAM for full model

Best for

  • Document‑processing workflows
  • Conversational assistants with review
  • Agentic tool‑calling systems

Who it’s for. Developers building conversational assistants or internal knowledge‑base systems that need robust German and English reasoning. The model is intended for use cases where outputs are reviewed before action, not for autonomous deployment.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—24.5 GB 25.8 GB
Too big Get file →
—45.7 GB 47.0 GB
Too big Get file →
—83.1 GB 84.5 GB
Too big Get file →

For a 24.46 GB file (Q8) the estimated VRAM at 8 k context is 25.8 GB. If 8 GB VRAM is available, use the 45.69 GB (Q4) file, which needs about 47 GB. For 8 GB VRAM the 83.14 GB (Q2) file is still too large; consider reducing context or using a smaller model.

What GPU do you need for Kolibri-1?

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: no quant fits at an 8K context.
  • 32 GB VRAM: the best quant that fits is the main file (24.5 GB file, about 25.8 GB in use).
  • 48 GB VRAM: the best quant that fits is the main file (45.7 GB file, about 47.0 GB in use).

Which Kolibri-1 quant should you choose?

If memory is tight, the smallest file is the first one in the table at 24.5 GB, with the quality trade-offs that come with fewer bits.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturekolibri1
Parameters78.1B (approximate, mixture of experts)
Layers50
Trained context262,144 tokens
KV cache (fp16)100 KB per token

Source

Quantized byaparusel
FamilyKolibri
Licenseapache-2.0
Downloads (HF)903
Last updated2026-10-06

Frequently asked questions about Kolibri-1

How much VRAM does Kolibri-1 need?

About 47.0 GB at its main file with an 8K context window. That is the 45.7 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 Kolibri-1 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 25.8 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Kolibri-1 GGUF?

It depends on the quant: file is 24.5 GB, file is 45.7 GB, file is 83.1 GB.

What context length does Kolibri-1 support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.4 GB at 32K.

Where can I download Kolibri-1 GGUF?

The files are on Hugging Face in the aparusel/ds4-kolibri-1-gguf repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Kolibri-1 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 Kolibri-1 on 8 GB VRAM?

No, the smallest GGUF file requires at least 47 GB VRAM at 8 k context. You would need more VRAM or a smaller model.

Does Kolibri-1 support CUDA?

No. Kolibri-1 runs on Metal or the CPU reference backend; CUDA, ROCm, and vision are not supported.

What license applies to Kolibri-1?

The model is released under the Apache‑2.0 license, with attribution required for redistributed quantized copies.

Other GGUFs of Kolibri-1

Similar models

Browse all →

More Kolibri models

All Kolibri →

Other large models

All large models →