K2 Horizon 7B

7B · Family: K2 Horizon

Dense 7B model (7B-core, 9B total) from IFM’s K2 Horizon family: 512K context, Apache 2.0, fully open (LLM360), with reasoning and tool use.

⏳ Preview✓ Public access⚖ Open sourceLLMReasoning model📁 K2 Horizon

Context window

512K

tokens

Parameters

7B (dense, decoder-only; ~9B total)

parameters

Release date

3 September 2026

🏢Institute of Foundation ModelsProducer

Access:DownloadDeployment:💻 Local☁ Cloud

Overview

K2 Horizon 7B is a dense, decoder-only language model from the K2 Horizon family developed by the Institute of Foundation Models (IFM). It is the smallest of the family’s larger dense variants — a 7B-core model (about 9B total parameters) in which all parameters are activated per token. Unlike the family’s sparse Mixture-of-Experts variants, it does not use expert routing.

The model supports a native 512K context window (524,288 tokens). By default it runs with high reasoning effort, returning its reasoning trace separately from the answer, supports tool calling (k2_horizon parser), and targets reasoning, coding, scientific, and agentic tasks.

K2 Horizon 7B is fully open following the LLM360 principle — IFM releases the training data and recipe, training code, and evaluation resources. Weights are published under Apache 2.0 in safetensors (BF16) format. Serving was validated on a single GPU with tensor parallelism (TP=1) via vLLM and SGLang (with FlashAttention-3); at least 32,768 output tokens are recommended, with sampling temperature 1.0, top_p 0.95.

Classification

LLMReasoning model

Family: K2 Horizon

Access & deployment

Download

LocalCloud

Weights: Open source

Key parameters

📏 Context: 512K

🧩 Parameters: 7B (dense, decoder-only; ~9B total)

✓ Tools · ✓ Fine-tuning

📥 Input: text

Technical specification

Context window

512K

tokens

Parameters

7B (dense, decoder-only; ~9B total)

parameters

License

Apache 2.0

Hardware requirements

BF16 serving on a single GPU with tensor parallelism (TP=1), e.g. via vLLM or SGLang (with FlashAttention-3). At least 32,768 output tokens recommended; sampling: temperature 1.0, top_p 0.95.

Features:✓ Tool use✓ Fine-tuning

Modalities

⬇ Input

text

⬆ Output

textcode

Capabilities and applications

Native model capabilities

Reasoning

The model's ability to reason logically and solve complex problems.

Category: reasoning

Coding

Generating, analysing and modifying code in many programming languages. Covers writing functions, debugging, refactoring, code review, and creating tests. Measured by benchmarks such as HumanEval and SWE-bench.

Category: coding

Tool use

The model's ability to call external functions, APIs and tools during a conversation: calculator, search engine, code editor, database. The model decides when and how to use a tool and interprets its result.

Category: planning

Long context

Support for large context windows — tens to hundreds of thousands (or millions) of input tokens. Enables analysis of entire codebases, long documents, and many parallel conversations without losing earlier information. GPT-5.1 supports 400,000 tokens.

Category: language

Adaptive reasoning effort

The model decides how much 'thinking' to allocate to a given query: simple questions are answered quickly, complex problems receive more inference cycles. A GPT-5.1 feature (both Instant and Thinking) that shortens time on easy tasks and extends it for hard ones.

Category: reasoning