Aiqre

Model library / danish-foundation-models

DFM-Mimir

DFM-Mimir is a 1.8 billion-parameter HRM-Text causal language model aimed at English and Danish tasks.

1.8B

Parameters in total

4,096

Context length in tokens, as published

3 GiB

Weights at the released precision

apache-2.0

Licence declared on the repository

Overview

What this model is.

Mimir is a Hierarchical Reasoning Model trained from scratch on permissible data for English and Danish, with a focus on open-source and ethically sourced datasets. The developers claim it achieves competitive results against larger frontier models on a range of benchmarks, and it is intended for research and application in those two languages.

In practice the model contains 1.8 billion parameters and accepts up to 4,096 tokens per prompt. It is released under the Apache-2.0 licence, with weights not publicly published and no expert layers in its architecture. The model type is hrm_text and it uses the HrmTextForCausalLM architecture.

Specification

The published shape.

Repositorydanish-foundation-models/DFM-Mimir
Publisherdanish-foundation-models
Published3 August 2026
ArchitectureHrmTextForCausalLM
Model typehrm_text
Parameters1,786,775,040
Layers16
Hidden size1,536
Attention heads12, 12 key/value heads
Context length4,096 tokens
Vocabulary262,144 tokens
Licenceapache-2.0

Memory

How much VRAM the weights need.

Parameter count multiplied by the bytes each format uses. The key-value cache comes on top.

PrecisionWeightsNotes
bf163.3 GiBFull precision as released
fp81.7 GiB8-bit, near-lossless on most models
int40.8 GiB4-bit, smallest footprint

The cache costs 96 KiB per token at 16-bit, so the full 4,096-token context of one request needs about 0.4 GiB. Concurrency multiplies that number, not the weights.

Hardware

Which card runs it.

At the released precision, with 90% of the card given to the server and 2.5 GiB kept for the runtime. The context column is what is left for the cache on that setup.

GPUCards neededFree for cacheContext it holds
RTX 4090 24GB 1 15.8 GiB the full 4,096 tokens
RTX 5090 32GB 1 23.0 GiB the full 4,096 tokens
L40S 48GB 1 37.4 GiB the full 4,096 tokens
A100 80GB 1 66.2 GiB the full 4,096 tokens
H100 80GB 1 66.2 GiB the full 4,096 tokens
RTX PRO 6000 96GB 1 80.6 GiB the full 4,096 tokens
H200 141GB 1 121.1 GiB the full 4,096 tokens

Serving

Running it yourself.

A vLLM launch line for the shape above. Check the model card for a runtime the publisher recommends.

vllm serve danish-foundation-models/DFM-Mimir \
  --max-model-len 4096 \
  --gpu-memory-utilization 0.90

Questions

The things people ask about this model.

How much GPU memory does DFM-Mimir need?

The weights take about 3 GiB at the released precision. That fits on one RTX 4090 24GB, which leaves roughly 16 GiB for the key-value cache.

What licence does DFM-Mimir use?

The repository declares apache-2.0. Read the licence text before commercial use: the name alone does not tell you what is allowed.

How long a context does it support?

The configuration allows 4,096 tokens. Whether the whole window is usable depends on the memory left for the cache, which the hardware table works out per card.

Can I use it through an API instead of hosting it?

This page is a reference, and Aiqre does not serve this model today. We deploy open-weight models on dedicated EU hardware on request, so if you want an OpenAI-compatible endpoint for it, email [email protected] with the model name and roughly what volume you expect.

Also in the library

Related models.

openbmb/MiniCPM5-2B

MiniCPM5-2B is a 2-billion-parameter dense transformer aimed at on-device and resource-constrained deployments.

IFM/K2-Horizon-0.9B

K2-Horizon-0.9B is a 1.1 billion-parameter dense decoder-only model with a 128 k token context window.

openbmb/MiniCPM5-2B-Midtrain

MiniCPM5-2B is a 2.5-billion-parameter dense Llama-style causal language model released by openbmb.

menezesbruno/manaca-1b-base

Manacá-1B is a 1.7 billion-parameter Llama-style decoder-only model pretrained for Brazilian Portuguese.

Hosting

Want this model on a dedicated EU GPU?

Tell us the model and roughly what volume you expect. We reply with a price and an OpenAI-compatible endpoint, with your prompts never stored.