K2-Horizon-MoVA-36B-A4B-GGUF

[!NOTE]

This repository contains GGUF versions of the IFM/K2-Horizon-MoVA-36B-A4B for use with llama.cpp.

The model tensors are stored in their original BF16 precision. The GGUF files include the tokenizer metadata and a llama.cpp-compatible chat template.

Compatibility: These models require a version of llama.cpp containing K2 Horizon architecture support. PR to llama.cpp is in progress. MBZUAI-IFM fork of llama.cpp is in https://github.com/MBZUAI-IFM/llama.cpp/tree/model/K2Horizon

K2-Horizon-MoVA-36B-A4B is the sparse member of the K2-Horizon family: a Mixture-of-Experts model with Mixture-of-Values attention (MoVA) that stores 36B parameters and runs 4B per token. We have released the final checkpoint; intermediate checkpoints, along with the data and the training code, will be released.

K2-Horizon-MoVA-36B-A4B Highlights

Frontier-class results at 4B active parameters. On agentic and reasoning benchmarks it outscores open weight dense (approximately 30B model size) and MoE models up to 15× its size; and also performs competitively against closed frontier models (see Benchmark Results).
512K context. Native 524,288-token context from the midtraining stages onward.
Intermediate checkpoints. Intermediate checkpoints will be released so capability changes can be studied across training rather than at a single checkpoint.
Fully open. Training data/recipe and the training code will be made public.

Benchmark Results

Open-weight models

K2-Horizon-MoVA-36B-A4BNemotron 3 UltraNemotron 3 SuperG9v3-39A5BQwen3.6-35B-A3BMuse Glimmer-30BGemma 4 31B-it

Params36B550B120B39B35B30B31B

Activated params4B55B12B5B3B30B31B

ArchitectureMoEMoEMoEMoEMoEDenseDense

Agents

tau3-Banking

Agentic tool use

26.814.210.322.19.323.514.8

Coding

Terminal-Bench 2.1

Agentic terminal use

58.653.938.632.644.951.743.4

SciCode

Scientific coding

38.939.936.034.035.843.643.4

Scientific Reasoning

Humanity's Last Exam (without tools)

Expert-level reasoning

25.228.420.817.522.222.023.6

GPQA Diamond

Graduate-level science QA

80.886.780.080.584.183.585.7

CritPt

Frontier physics reasoning

2.13.13.10.30.32.61.4

General

AA-LCR

Long-context reasoning

66.371.060.362.066.780.068.3

AA-Omniscience Accuracy

Factual accuracy

18.822.624.314.918.827.020.0

AA-Omniscience Non-Hallucination

Non-hallucination rate

69.270.313.087.049.518.115.0

Scores in %. Bold marks the best score in each row. Sections follow the Artificial Analysis Intelligence Index categories. Baseline scores are from Artificial Analysis; Muse Glimmer-30B at high reasoning effort, all other open models in their reasoning mode.

Quickstart

Serving

vLLM, recipe at recipes.vllm.ai/IFM:

vllm serve IFM/K2-Horizon-MoVA-36B-A4B \
  --revision main \
  --tensor-parallel-size 2 \
  --enable-expert-parallel \
  --trust-remote-code \
  --dtype bfloat16 \
  --max-model-len 131072 \
  --reasoning-parser k2_horizon \
  --tool-call-parser k2_horizon \
  --enable-auto-tool-choice

SGLang recipe validated on 2× H200 in the SGLang K2 Horizon cookbook:

python3 -m sglang.launch_server \
  --model-path IFM/K2-Horizon-MoVA-36B-A4B \
  --revision main \
  --tp 2 \
  --ep 2 \
  --dtype bfloat16 \
  --attention-backend fa3 \
  --json-model-override-args '{"xllm_source_router_gemm_partitions":2}' \
  --reasoning-parser k2_horizon \
  --tool-call-parser k2_horizon \
  --host 0.0.0.0 --port 30000

API Usage

[!Tip]

Recommended settings: reasoning_effort="high", temperature=1.0, top_p=0.95.

Reasoning depth is selected per request through chat_template_kwargs. Thinking is returned in reasoning_content and the answer in content.

from openai import OpenAI

client = OpenAI(base_url="http://localhost:30000/v1", api_key="EMPTY")
response = client.chat.completions.create(
    model="IFM/K2-Horizon-MoVA-36B-A4B",
    messages=[{"role": "user", "content": "Explain the result step by step."}],
    temperature=1.0,
    top_p=0.95,
    max_tokens=32768,
    extra_body={"chat_template_kwargs": {"reasoning_effort": "high", "tool_call_format": "xml"}},
)
message = response.choices[0].message
print("Reasoning:", getattr(message, "reasoning_content", None))
print("Answer:", message.content)

Our model supports multiple tool calls formats, which can be changed with chat_template_kwargs. The supported values are json, xml, and xml_typed . The default is xml. Keep --tool-call-parser k2_horizon enabled to parse the selected format.

Transformers

Validated with Transformers 5.15.0, PyTorch 2.13.0, Safetensors 0.8.0.

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "IFM/K2-Horizon-MoVA-36B-A4B"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id, device_map="auto", dtype="bfloat16", low_cpu_mem_usage=True, trust_remote_code=True
)

inputs = tokenizer("Explain why long-context evaluation is difficult.", return_tensors="pt").to(model.device)
inputs.pop("token_type_ids", None)
outputs = model.generate(**inputs, max_new_tokens=32768, temperature=1.0, top_p=0.95, do_sample=True)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Best Practices

1. Reasoning effort: always high. All reported results use high reasoning effort. Pass {"chat_template_kwargs": {"reasoning_effort": "high"}} on every request.

2. Sampling parameters. temperature=1.0, top_p=0.95.

3. Serving. Use the validated SGLang recipe above: BF16, TP=2, FlashAttention-3, and the xllm_source_router_gemm_partitions override, which preserves the checkpoint's router numerics. Full recipes for every K2-Horizon size, with measured H200 latency and throughput, are in the SGLang cookbook and the vLLM recipe.

4. Parsers. Enable the k2_horizon reasoning parser for chat, and add the k2_horizon tool-call parser for agent use. Leave both off for plain completion-style generation.

Citation

@misc{k2horizon2026,
  title  = {Introducing K2 Horizon: Frontier Performance, Radically Open},
  author = {{IFM Team}},
  year   = {2026},
  url    = {https://ifm.ai/blog/k2/},
}

IFM/K2-Horizon-MoVA-36B-A4B-GGUF