--- title: K2-Horizon-3.7B canonical_url: "https://www.modelscope.cn/models/IFM/K2-Horizon-3.7B" md_url: "https://www.modelscope.cn/models/IFM/K2-Horizon-3.7B.md" repository: IFM/K2-Horizon-3.7B last_updated: 2026-09-07 license: apache-2.0 pipeline_tag: text-generation tasks: - text-generation model_type: - k2_horizon architectures: - K2HorizonForCausalLM parameters: 5.1B tensor_type: - BF16 library_name: - transformer - safetensors language: - en downloads: 34 stars: 0 tags: - k2-horizon - 3.7b - dense - open-weights - ifm --- # K2-Horizon-3.7B > K2-Horizon-3.7B - IFM 在 ModelScope 开源的模型。K2-Horizon-3.7B is the small dense member of the K2-Horizon family: a 3.7B-core decoder-only model with a 512K context window. IFM/K2-Horizon-3.7B 是 ModelScope 魔搭社区上的 5.1B 参数text-generation模型，采用 apache-2.0 许可。 - **Repository**: IFM/K2-Horizon-3.7B - **License**: apache-2.0 - **Tasks**: text-generation - **Parameters**: 5.1B - **Tags**: k2-horizon, 3.7b, dense, open-weights, ifm - **Downloads**: 34 - **Stars**: 0 - **Last updated**: 2026-09-07 Source: https://www.modelscope.cn/models/IFM/K2-Horizon-3.7B --- # K2-Horizon-3.7B K2-Horizon-3.7B is the small dense member of the K2-Horizon family: a 3.7B-core decoder-only model with a 512K context window.

K2-Horizon-3.7B benchmark results

## K2-Horizon-3.7B Highlights - **Strong small-model baseline.** A dense model evaluated on the same agentic, coding, and reasoning benchmarks as the rest of the family. - **512K context.** Native 524,288-token context from the midtraining stages onward. - **Intermediate checkpoints.** Intermediate checkpoints are released so capability changes can be studied across training rather than at a single checkpoint. - **Fully open.** Training data and recipe, training code, and evaluation resources are public. ## Benchmark Results The chart at the top of this card shows K2-Horizon-3.7B against selected reference models. The table below lists every comparison model used in the figure. ### Full Results
Open-weight dense models
K2-Horizon-3.7BQwen3.5-4BG9v3-3BGranite 4.2-3BNemotron 3 Nano-4B
# Params3.7B4B3B3B4B
# Activated params3.7B4B3B3B4B
ArchitectureDenseDenseDenseDenseDense
Math
HMMT Feb 2026
Competition mathematics
70.561.634.157.234.7
Coding
SWE-bench Verified
Software engineering
68.641.216.432.21.8
Scientific Reasoning
GPQA Diamond
Graduate-level science QA
65.477.143.855.951.3
HLE
Expert-level reasoning
12.99.94.56.64.9
Coding
SciCode
Scientific coding
25.916.117.724.916.4
Terminal-Bench 2.1
Agentic terminal use
25.125.86.013.93.7
Agents
tau3-Banking
Agentic tool use
17.76.8—5.6—
BFCL v4
Function calling
50.955.747.950.836.8
Scores in %. Bold marks the best score in each row. Baseline protocols may differ; ## Quickstart ### Serving vLLM, recipe at [recipes.vllm.ai/IFM](https://recipes.vllm.ai/IFM): ```shell vllm serve IFM/K2-Horizon-3.7B \ --trust-remote-code \ --dtype bfloat16 \ --max-model-len 131072 \ --tensor-parallel-size 1 \ --reasoning-parser k2_horizon \ --enable-auto-tool-choice \ --tool-call-parser k2_horizon ``` SGLang, this is the recipe validated in the [SGLang K2 Horizon cookbook](https://docs.sglang.io/cookbook/autoregressive/IFM/K2-Horizon): ```shell sglang serve \ --model-path IFM/K2-Horizon-3.7B \ --revision c177771836a4c460743c00002c22483f6f18d1eb \ --tp 1 \ --dtype bfloat16 \ --attention-backend fa3 \ --reasoning-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`, and at least 32,768 output tokens. > Reasoning depth is selected per request through `chat_template_kwargs`. Thinking is returned in `reasoning_content` and the answer in `content`. ```python from openai import OpenAI client = OpenAI(base_url="http://localhost:30000/v1", api_key="EMPTY") response = client.chat.completions.create( model="IFM/K2-Horizon-3.7B", 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"}}, ) message = response.choices[0].message print("Reasoning:", getattr(message, "reasoning_content", None)) print("Answer:", message.content) ``` ### Transformers Validated with Transformers 5.15.0, PyTorch 2.13.0, Safetensors 0.8.0. ```python from transformers import AutoModelForCausalLM, AutoTokenizer model_id = "IFM/K2-Horizon-3.7B" 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; `medium` and `low` trade accuracy for speed and are not recommended for evaluation. 2. **Sampling parameters.** `temperature=1.0`, `top_p=0.95`. 3. **Output length.** Allow at least 32,768 output tokens so reasoning is never cut off. Truncated reasoning is a failed response, not a shorter one. 4. **Serving.** Use the validated SGLang recipe above: BF16, TP=1, FlashAttention-3. Full recipes for every K2-Horizon size, with measured H200 latency and throughput, are in the [SGLang cookbook](https://docs.sglang.io/cookbook/autoregressive/IFM/K2-Horizon). 5. **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. 6. **Revisions.** Pin a revision tag when reproducibility matters. `main` is the default checkpoint; `base_final` and the `mid_*_final` tags identify training stages. ## Citation ```bibtex @misc{k2horizon2026, title = {Introducing K2 Horizon: Frontier Performance, Radically Open}, author = {{IFM Team}}, year = {2026}, url = {https://ifm.ai/blog/k2/}, } ```