K2 Horizon 7B
An open-source language model with 7 billion parameters, known for its cost-effectiveness. It achieves performance on the Artificial Analysis Intelligence Index approaching that of larger-scale models, is released in GGUF format on Hugging Face, and is suited for local deployment on consumer-grade hardware.
Timeline (last 90 days)
K2 Horizon 7B is a small-scale language model with 7 billion parameters.
K2 Horizon 7B achieved a score on the Artificial Analysis Intelligence Index that falls between Qwen 3.6 27B and Qwen 3.6 35BA3b.
K2 Horizon 7B has been released on Hugging Face in GGUF format, in the repository IFM/K2-Horizon-7B-GGUF.
A poster used K2 Horizon 7B to compile the latest version of llama.cpp for CUDA and reported that the model performed well.
Compared to flagship models with tens of billions of parameters, 7B-level models have advantages in deployment cost, response latency, and privacy controllability; running locally means data never leaves the device and no API call fees need to be paid.
All Facts (3)
K2 Horizon 7B is a small-scale language model with 7 billion parameters.
50%UnverifiedCompared to flagship models with tens of billions of parameters, 7B-level models have advantages in deployment cost, response latency, and privacy controllability; running locally means data never leaves the device and no API call fees need to be paid.
50%UnverifiedK2 Horizon 7B has been released on Hugging Face in GGUF format, in the repository IFM/K2-Horizon-7B-GGUF.
50%