The AI arena is free today

Open Superagent
Institute of Foundation ModelsReleased on Sep 3, 2026

K2 Horizon 3.7B: Benchmarks, Pricing & Context Window

K2 Horizon 3.7B is a language model from Institute of Foundation Models, released in September 2026.

K2 Horizon 3.7B is the small dense member of IFM's (MBZUAI) fully open K2 Horizon family: a 3.7B-core decoder-only reasoning model (36 layers; about 5.1B parameters in the released safetensors including its 250K-token vocabulary

K2 Horizon 3.7B benchmarks

Capability tiers

Standing within each category, adjusted for leaderboard depth.

Real tasks performance

High-confidence performance for K2 Horizon 3.7B across real-world prompt categories. Only 95% intervals at most 4 points wide are shown.

Performance by conversation depth

How K2 Horizon 3.7B holds up as conversations get longer.

Quality Tracker

K2 Horizon 3.7B Performance Across Datasets

Scores sourced from the model's scorecard, paper, or official blog posts

LLM Stats Logollm-stats.com - Fri Oct 30 2026
Notice missing or incorrect data?

K2 Horizon 3.7B model size

K2 Horizon 3.7B has 3.7 billion parameters and was trained on 22.9 trillion tokens. See how it compares to other models in the same parameter range.

ParametersTraining tokens
3.7B
22.9Ttokens
6189× tokens-to-params ratio
Small (3–10B)
3.7B
1B7B70B405B

Try now

huggle
K2 Horizon 3.7Bin Huggle

Make it with
K2 Horizon 3.7B.

K2 Horizon 3.7B

K2 Horizon 3.7B latency

K2 Horizon 3.7B time to first token, sustained output throughput, and failed-request rate from live model usage over the trailing 7 days.

K2 Horizon 3.7B examples

Recent arena outputs from K2 Horizon 3.7B, picked from the highest-ranked matchups.

K2 Horizon 3.7B license

K2 Horizon 3.7B is released under the Apache 2.0 license, which permits commercial use, has 3.7B parameters.

License
Apache 2.0
Commercial use allowed
Parameters
3.7B

Apache License 2.0 - allows commercial use

K2 Horizon 3.7B resources

Official sources for K2 Horizon 3.7B: official launch post, source repository, model weights.

K2 Horizon 3.7B vs other models

The most-compared alternatives to K2 Horizon 3.7B are Phi 4 Reasoning Plus, Qwen3 VL 32B Instruct, Nova 2 Lite. Open any pair side-by-side for benchmarks, pricing, context, and latency.

Models like K2 Horizon 3.7B

Models ranked just above and below K2 Horizon 3.7B by LLM Stats score.

 

Phi 4 Reasoning Plus

Score pending
 

Qwen3 VL 32B Instruct

Score pending
 

Nova 2 Lite

Score pending
 

Muse Glimmer-30B

Score pending
 

Kimi K2-Instruct-0905

Score pending
 

GLM-4.5

Score pending

FAQ

Common questions about K2 Horizon 3.7B.

When was K2 Horizon 3.7B released?

K2 Horizon 3.7B was released on September 3, 2026 by ifm. This is the official K2 Horizon 3.7B release date tracked on LLM Stats.

How big is K2 Horizon 3.7B?

K2 Horizon 3.7B has 3.7 billion parameters. It was trained on 22.9 trillion tokens. It ships as an open-weight model, so you can download and run it on your own hardware.

Who created K2 Horizon 3.7B?

K2 Horizon 3.7B was created by ifm.

What is the license for K2 Horizon 3.7B?

K2 Horizon 3.7B is released under the Apache 2.0 license. This is an open-source / open-weight license that permits self-hosting.

Where is the K2 Horizon 3.7B paper or technical report?

K2 Horizon 3.7B has a paper or technical report available at https://ifm.ai/blog/k2/. Use that source for architecture, training, release and evaluation details.

What models should I compare K2 Horizon 3.7B against?

Common K2 Horizon 3.7B comparisons include K2 Horizon 3.7B vs Phi 4 Reasoning Plus, K2 Horizon 3.7B vs Qwen3 VL 32B Instruct, K2 Horizon 3.7B vs Nova 2 Lite. Compare them side by side for benchmark scores, pricing, context window, latency and provider availability.