whatmodel.sh / K2 Horizon 7B

K2 Horizon 7B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

K2 Horizon 7B is an open-weights language model from Institute of Foundation Models (MBZUAI), released September 2026. Its published size is 9B parameters and a 524k-token context window. Licence: apache-2.0.

No comparable score yet. Its published results are below, but they don't overlap enough with other models to calculate a score.

$ how it compares

Scored LLMs released closest to K2 Horizon 7B, out of 1132 with a score.

#modelscorewhereRAMreleased
1GPT-6 Astra OpenAI87±1api $20.00/M2026-09
2Claude Opus 5.5 Anthropic87±1api $8.00/M2026-09
3GPT-6.1 Sol OpenAI85±1api $4.00/M2026-09
4Claude Fable 5.1 Anthropic85±1api $20.00/M2026-09
5Claude Sonnet 5.5 Anthropic84±1api $4.00/M2026-09

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasksnot ranked on this task yet · 4 tests
testresultvs the others in its rundifficultycountswhy
SWE-bench Verified lab70.6% ✎▁▁▁▁▁▁▂▄█▆▁▁ beats 78/131—not counted yet: no other model in the same run to compare it with
Terminal-Bench 2.1 lab39.1% ✎▁·▁·▁▁▄▃▄█▃▇ beats 14/56—not counted yet: no other model in the same run to compare it with
τ³-bench Banking lab25.8% ✎▂▂·▅█▂▂▃▂▅▃▃ beats 18/20—not counted yet: no other model in the same run to compare it with
BrowseComp with context management lab59.0% ✎▅·▅▃·█▃·█▅·▃ beats 1/14—not counted yet: no other model in the same run to compare it with
codingnot ranked on this task yet · 2 tests
testresultvs the others in its rundifficultycountswhy
SWE-bench Verified lab70.6% ✎▁▁▁▁▁▁▂▄█▆▁▁ beats 78/131—not counted yet: no other model in the same run to compare it with
SciCode (level unspecified) lab31.6% ✎▁····▃▄▄▄█▆▂ beats 13/34—not counted yet: no other model in the same run to compare it with
mathnot ranked on this task yet · 1 test
testresultvs the others in its rundifficultycountswhy
HMMT February 2026 lab73.3% ✎▃▃▂▃·██▆█▅▃▅ beats 15/33—not counted yet: no other model in the same run to compare it with
web searchnot ranked on this task yet · 1 test
testresultvs the others in its rundifficultycountswhy
BrowseComp with context management lab59.0% ✎▅·▅▃·█▃·█▅·▃ beats 1/14—not counted yet: no other model in the same run to compare it with
other tests1 not in a task · 0.0% of its score
testresultvs the others in its rundifficultycountswhy
Humanity's Last Exam lab18.6% ✎▁▁▁▃▆▆▇██▆▁▃ beats 48/104—not counted yet: no other model in the same run to compare it with

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run K2 Horizon 7B · needs ~5.9 GB · from huggingface.co/IFM (official)
    chat in the terminal
    llama cli -hf IFM/K2-Horizon-7B-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf IFM/K2-Horizon-7B-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run K2 Horizon 7B · needs ~5.9 GB · from huggingface.co/IFM (official)

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/IFM/K2-Horizon-7B-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run K2 Horizon 7B · needs ~5.9 GB · from huggingface.co/IFM (official)
    chat in the terminal
    ollama run hf.co/IFM/K2-Horizon-7B-GGUF:Q4_K_M

Official weights and documentation: huggingface.co/IFM ↗

$ published results · 7 results, not enough overlap for a score
testresultsource and conditions
HMMT February 202673.3%IFM/K2-Horizon-7B model card · lab-reported

Own K2-Horizon-7B final release column. Best Practices states all reported results use high reasoning effort; temperature1.0, top_p0.95, at least32768 output tokens. Validation recipe BF16, TP1, FlashAttention3 is a serving recommendation, not an assumed separate evaluation run. HLE tool mode, SWE scaffold and SciCode main/subproblem level are not specified; retained without inference. No reference-model, AA-LCR or throughput values imported. Review date is not the evaluation-run date.

SWE-bench Verified70.6%IFM/K2-Horizon-7B model card · lab-reported

Own K2-Horizon-7B final release column. Best Practices states all reported results use high reasoning effort; temperature1.0, top_p0.95, at least32768 output tokens. Validation recipe BF16, TP1, FlashAttention3 is a serving recommendation, not an assumed separate evaluation run. HLE tool mode, SWE scaffold and SciCode main/subproblem level are not specified; retained without inference. No reference-model, AA-LCR or throughput values imported. Review date is not the evaluation-run date.

Humanity's Last Exam18.6%IFM/K2-Horizon-7B model card · lab-reported

Own K2-Horizon-7B final release column. Best Practices states all reported results use high reasoning effort; temperature1.0, top_p0.95, at least32768 output tokens. Validation recipe BF16, TP1, FlashAttention3 is a serving recommendation, not an assumed separate evaluation run. HLE tool mode, SWE scaffold and SciCode main/subproblem level are not specified; retained without inference. No reference-model, AA-LCR or throughput values imported. Review date is not the evaluation-run date.

SciCode (level unspecified)31.6%IFM/K2-Horizon-7B model card · lab-reported

Own K2-Horizon-7B final release column. Best Practices states all reported results use high reasoning effort; temperature1.0, top_p0.95, at least32768 output tokens. Validation recipe BF16, TP1, FlashAttention3 is a serving recommendation, not an assumed separate evaluation run. HLE tool mode, SWE scaffold and SciCode main/subproblem level are not specified; retained without inference. No reference-model, AA-LCR or throughput values imported. Review date is not the evaluation-run date.

Terminal-Bench 2.139.1%IFM/K2-Horizon-7B model card · lab-reported

Own K2-Horizon-7B final release column. Best Practices states all reported results use high reasoning effort; temperature1.0, top_p0.95, at least32768 output tokens. Validation recipe BF16, TP1, FlashAttention3 is a serving recommendation, not an assumed separate evaluation run. HLE tool mode, SWE scaffold and SciCode main/subproblem level are not specified; retained without inference. No reference-model, AA-LCR or throughput values imported. Review date is not the evaluation-run date.

BrowseComp with context management59.0%IFM/K2-Horizon-7B model card · lab-reported

IFM own evaluation in the pinned K2-Horizon-7B model card, Full Results table, own column (reference-model columns not used). All reported results use high reasoning effort. BrowseComp uses the Discard-all@95k context-length protocol from the DeepSeek-V3.2 report. LCR (long-context reasoning, AA-LCR style) not used. Column 'K2-Horizon-7B', row 'BrowseCompWeb browsing', section 'Agents'.

τ³-bench Banking25.8%IFM/K2-Horizon-7B model card · lab-reported

IFM own evaluation in the pinned K2-Horizon-7B model card, Full Results table, own column (reference-model columns not used). All reported results use high reasoning effort. BrowseComp uses the Discard-all@95k context-length protocol from the DeepSeek-V3.2 report. LCR (long-context reasoning, AA-LCR style) not used. Column 'K2-Horizon-7B', row 'tau3-BankingAgentic tool use', section 'Agents'.

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json