LL

Mellum2 12B A2.5B Instruct ik llama

LlamaGGUF by ji-farthingUpdated 2026-06-04
Download on Hugging Face
Parameters12.1B
Context128K
Layers28
Architecturemellum
Smallest quant12.9 GB
Quick answer

Mellum2 12B A2.5B Instruct ik llama needs about 13.9 GB of memory at Q8_0 (a 12.9 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 2 quants, from 12.9 GB (Q8_0) to 24.3 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 1.9 GB.

Mellum2 12B A2.5B Instruct – a Mixture‑of‑Experts Llama model with 131,072 token context

Mellum2 12B A2.5B Instruct is a 12.1B parameter Llama model with a 2.5B active MoE parameter set. It supports a 131,072 token context length and is fine‑tuned for direct instruction following and code assistance.

Good at

  • Direct instruction following
  • Code assistance
  • Tool‑use support

Limits

  • Requires 8,000+ token context
  • High VRAM usage for Q8_0
  • No vision capabilities

Best for

  • Interactive chat agents
  • Coding assistants
  • Tool‑integrated workflows

Who it’s for. Developers and researchers needing a large Llama model with extensive context and Mixture‑of‑Experts architecture for instruction‑driven tasks. The model is released under Apache 2.0 and can be run with llama.cpp or ik_llama.cpp.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q8_012.9 GB 13.9 GB
Too big Get file →
BF1624.3 GB 25.3 GB
Too big Get file →

For 8 GB VRAM, the Q8_0 quant is too large – it requires about 13.9 GB at 8 k context. The BF16 quant is larger still, about 25.3 GB. If you have 16 GB or more, Q8_0 is the most efficient choice. For 24 GB or more, BF16 provides higher precision.

What GPU do you need for Mellum2 12B A2.5B Instruct ik llama?

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: the best quant that fits is Q8_0 (12.9 GB file, about 13.9 GB in use).
  • 24 GB VRAM: the best quant that fits is Q8_0 (12.9 GB file, about 13.9 GB in use).
  • 32 GB VRAM: the best quant that fits is BF16 (24.3 GB file, about 25.3 GB in use).

Which Mellum2 12B A2.5B Instruct ik llama quant should you choose?

Q8_0 (12.9 GB) is the middle option in this repository. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturemellum
Parameters12.1B (approximate, mixture of experts)
Active parameters2.5B
Layers28
Trained context131,072 tokens
KV cache (fp16)56 KB per token

Source

Quantized byji-farthing
FamilyLlama
Licenseapache-2.0
Downloads (HF)55
Last updated2026-06-04

Frequently asked questions about Mellum2 12B A2.5B Instruct ik llama

How much VRAM does Mellum2 12B A2.5B Instruct ik llama need?

About 13.9 GB at Q8_0 with an 8K context window. That is the 12.9 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 Mellum2 12B A2.5B Instruct ik llama on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 13.9 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Mellum2 12B A2.5B Instruct ik llama GGUF?

It depends on the quant: Q8_0 is 12.9 GB, BF16 is 24.3 GB.

What context length does Mellum2 12B A2.5B Instruct ik llama support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.9 GB at 32K.

Where can I download Mellum2 12B A2.5B Instruct ik llama GGUF?

The files are on Hugging Face in the ji-farthing/Mellum2-12B-A2.5B-Instruct-ik-llama-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 Mellum2 12B A2.5B Instruct ik llama 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 this model on 8 GB VRAM?

No. The smallest quant requires about 13.9 GB of VRAM at 8 k context.

Does the model support vision tasks?

No, the model is text‑only.

What license applies?

Apache 2.0, allowing commercial use and modification.

Other GGUFs of Mellum2-12B-A2.5B-Instruct

Similar models

Browse all →

More from ji-farthing

All by ji-farthing →

More Llama models

All Llama →

Other medium models

All medium models →