K2 Horizon MoVA 36B A4B
IFMapache-2.0text

K2 Horizon MoVA 36B A4B

Updated Sep 17, 2026
Parameters
37.4B
Context
524,288
License
apache-2.0
Updated
Sep 17, 2026

Intelligence benchmarks

Artificial Analysis indexes - compared with the best open and proprietary models

Intelligence

25.7

AA Index

Intelligence Index - K2 Horizon MoVA 36B A4B vs. the field

Best open-weight models (you can run locally) and leading proprietary models for context.

Claude Fable 5.1 (max with fallback)
53.4
closed
GPT-6 Astra (max)
52.8
closed
Claude Opus 5 (max)
50.7
closed
Muse Spark 1.3 (max)
48.2
closed
GPT-5.6 Sol (max)
47.1
closed
GLM-5.3
44.9
open
Kimi K3
43.8
open
K2 Horizon MoVA 36B A4B
25.7
open

Benchmark data from Artificial Analysis · updated 2026-09-17.

Standard benchmarks

Performance across standard evaluations

BenchmarkScore
GPQA82.2

Will it run on your hardware?

Pick your GPU memory - see which quantizations fit, and the cheapest card for the rest

Too big for 24 GB at any quant
0 of 4 quantizations fit K2 Horizon MoVA 36B A4B with real runtime overhead.

Need an exact figure for your context length? Use the VRAM calculator.

Run it locally

Copy-paste - running in under a minute

vLLMOpenAI-compatible API
vllm serve IFM/K2-Horizon-MoVA-36B-A4B

New to this? Start with Ollama · serve to many users with vLLM.

Deep dive

Notes, sources, and the full write-up

K2 Horizon MoVA 36B A4B is a 37.4B-parameter apache-2.0 model from MBZUAI Institute of Foundation Models. It scores 25.7 on the Artificial Analysis Intelligence Index. At Q4_K_M it needs roughly 22 GB of VRAM, placing it in the 12–24 GB GPU hardware tier.

Benchmarks

Artificial Analysis Intelligence Index - K2 Horizon MoVA 36B A4B vs. leading closed models:

ModelIntelligenceCodingGPQA
K2 Horizon MoVA 36B A4B25.7-82.2
Claude Fable 5.1 (max with fallback)53.4-93.7
GPT-6 Astra (max)52.8-96.1
Claude Opus 5 (max)50.7-93.2
Claude Fable 5 (with fallback)49.7-92.6
Muse Spark 1.3 (max)48.2-93.5

Source: Artificial Analysis (2026-09-14).

Popularity

K2 Horizon MoVA 36B A4B has 5,996 downloads in the last month on HuggingFace and 301 likes.

Frequently asked

Quick answers to common questions

How much VRAM does K2 Horizon MoVA 36B A4B need?

K2 Horizon MoVA 36B A4B with 37.4B parameters needs approximately 22 GB at Q4_K_M quantization. Use our VRAM calculator for an exact estimate.

Is K2 Horizon MoVA 36B A4B better than other IFM models?

K2 Horizon MoVA 36B A4B has 37.4B parameters with 524,288 context - a strong choice for general use.

What license is K2 Horizon MoVA 36B A4B under?

K2 Horizon MoVA 36B A4B is released under the apache-2.0 license, making it suitable for most commercial and personal projects.

What hardware runs K2 Horizon MoVA 36B A4B well?

With 37.4B parameters, K2 Horizon MoVA 36B A4B requires adequate VRAM. High-end GPUs like the RTX 4090 (24GB), RTX 5090 (32GB), or Mac Studio with unified memory are good options. Check our hardware directory for specific recommendations.

What is the best quantization for K2 Horizon MoVA 36B A4B?

Q4_K_M is the recommended sweet spot - ~98% of FP16 quality at ~27% of the size. Q5_K_M (~27 GB) is an option if you have spare VRAM. Use our VRAM calculator to compare.

How long can K2 Horizon MoVA 36B A4B's context window handle?

K2 Horizon MoVA 36B A4B supports a 524,288-token context window - enough for very long documents, codebases, or multi-turn conversations. Real-world usable context may vary by implementation.

What models compete with K2 Horizon MoVA 36B A4B?

K2 Horizon MoVA 36B A4B competes with other 19B–56B. Browse our model directory for comparisons, benchmarks, and community reviews to find the best fit.

Nearby options

Similar models and compatible hardware by spec

Comments coming soon

Configure NEXT_PUBLIC_GISCUS_REPO_ID and NEXT_PUBLIC_GISCUS_CATEGORY_ID at giscus.app to enable.