Aiqre

Model library / IFM

K2-Horizon-3.7B

K2-Horizon-3.7B is a 3.7 billion-parameter dense decoder-only model with a 512 K token context, released openly by IFM.

5.1B

Parameters in total

524,288

Context length in tokens, as published

9 GiB

Weights at the released precision

apache-2.0

Licence declared on the repository

Overview

What this model is.

K2-Horizon-3.7B is a dense decoder-only language model with 3.7 billion parameters, released by IFM. The publisher describes it as a strong small-model baseline evaluated on agentic, coding and reasoning benchmarks, and notes that it provides a 512 K token context window and fully open training data and code.

In deployment the model occupies roughly 5.1 billion parameters stored in bfloat16 format, fitting within the memory limits of modern GPUs. It accepts up to 524,288 tokens per request, enabling long-range tasks, and is distributed under the Apache-2.0 licence, allowing unrestricted commercial and research use.

Specification

The published shape.

RepositoryIFM/K2-Horizon-3.7B
PublisherIFM
Published2 September 2026
ArchitectureK2HorizonForCausalLM
Model typek2_horizon
Parameters5,058,255,360
Layers36
Hidden size2,560
Attention heads32, 8 key/value heads
Context length524,288 tokens
Vocabulary250,624 tokens
Weight formatbfloat16
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
bf169.4 GiBFull precision as released
fp84.7 GiB8-bit, near-lossless on most models
int42.4 GiB4-bit, smallest footprint

The cache costs 144 KiB per token at 16-bit, so the full 524,288-token context of one request needs about 72.0 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 9.7 GiB about 70,474 tokens
RTX 5090 32GB 1 16.9 GiB about 122,903 tokens
L40S 48GB 1 31.3 GiB about 227,761 tokens
A100 80GB 1 60.1 GiB about 437,476 tokens
H100 80GB 1 60.1 GiB about 437,476 tokens
RTX PRO 6000 96GB 1 74.5 GiB the full 524,288 tokens
H200 141GB 1 115.0 GiB the full 524,288 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 IFM/K2-Horizon-3.7B \
  --max-model-len 69632 \
  --gpu-memory-utilization 0.90

Questions

The things people ask about this model.

How much GPU memory does K2-Horizon-3.7B need?

The weights take about 9 GiB at the released precision. That fits on one RTX 4090 24GB, which leaves roughly 10 GiB for the key-value cache.

What licence does K2-Horizon-3.7B 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 524,288 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.

IFM/K2-Horizon-375B-A23B

K2-Horizon-375B-A23B is an open-weight sparse MiE language model from IFM with 379 B parameters, 23 B active per token and a 512 …

IFM/K2-Horizon-7B

K2-Horizon-7B is a 9-billion-parameter decoder-only language model with a 524k token context window, released under Apache-2.0.

IFM/K2-Horizon-32B

K2-Horizon-32B-Stage1 is a 32-billion-parameter dense decoder-only model released by IFM for benchmarking agentic, coding and rea…

IFM/K2-Horizon-MoVA-36B-A4B

K2-Horizon-MoVA-36B-A4B is a sparse Mixture-of-Experts language model with MoVA attention, released by IFM.

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.