IFM releases K2-Horizon-7B-Uno diffusion-augmented LLM, claims 5200 tps speedup
The Institute of Foundation Models (IFM) has posted a new 7‑billion‑parameter language model, K2‑Horizon‑7B‑Uno, on Hugging Face. The model adds a diffusion pathway to the existing K2‑Horizon‑7B architecture via LoRA‑based adapters, promising the same generation quality while running at roughly 5,200 tokens per second on benchmark tests. The repository contains only the adapters; the base‑model…
Key points
- K2‑Horizon‑7B‑Uno adds LoRA diffusion adapters to a 7B LLM, claiming 5,200 tps with unchanged quality.
- Model and adapters are hosted on Hugging Face; base weights are separate under IFM/K2‑Horizon‑7B.
- 46,557 downloads in the last month indicate strong early community interest.
Developers can load the model with PEFT by wrapping the base model in a PeftModel object, and the project provides example notebooks for Google Colab and Kaggle. Selected evaluation scripts are available, with a full benchmark suite slated for release later. The model has already attracted attention, recording 46,547 downloads in the past month. The authors ask users to cite the accompanying paper and project page when using the model.
If the performance claims hold, K2‑Horizon‑7B‑Uno could offer a practical route to faster inference for applications that need high‑throughput text generation without sacrificing quality, especially for developers already using PEFT‑based fine‑tuning pipelines.
Model page: K2-Horizon-7B-Uno →
The story so far
2 episodes →- IFM releases K2-Horizon-7B-Uno diffusion-augmented LLM, claims 5200 tps speedupthis story
IFM/K2-Horizon-7B-Uno · Hugging Face - 5200tps with no quality loss
huggingface.co · 11 September 2026
Loading the full article…
This text was published by huggingface.co. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Reddit discussionreddit.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Introducing Mistral Large 4 · 7 src
- OpenAI boosts GPT-6 speed by 50% in 28 Days update · 1 src
- Reflection AI debuts Beam, a 501B open-weight model rivaling GLM 5.2 with 3-4x less compute · 10 src
- Author tests four AI assistants on flawed forecasting scenarios · 1 src
- Anthropic prompts users to share voice data to enhance AI models · 8 src
Comments
via GitHub Discussions