whatmodel.sh / Apertus 8B Instruct (2509)

Apertus 8B Instruct (2509)

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

Apertus 8B Instruct (2509) is an open-weights language model from Swiss AI, released August 2025. Its published size is 8.05B parameters and a 66k-token context window. Licence: apache-2.0.

Catalogue identity notes

swiss-ai/Apertus-8B-Instruct-2509 @b946d40 (created 2025-08-13, apache-2.0). Swiss AI (EPFL, ETH Zurich, CSCS) Apertus 1.0 8B instruct model, September 2025. Independent rows: EuroEval English and Miðeind Icelandic leaderboard.

Model information sources

Release date ↗ · Weights licence ↗

HF safetensors metadata 8,053,338,176 parameters.

38.1±3.6#563 of 719 LLMsfrom 5 benchmark results across 3 sources

$ how it compares

The LLMs ranked either side of Apertus 8B Instruct (2509), out of 828 with a score.

#modelscorewhereRAMreleased
653Tulu 2 13B Allen Institute for AI38±2local~9.8GB2023-11
654Phi-3-mini-4k-instruct (Apr 2024) Microsoft38±2local~4GB2024-04
655Apertus 8B Instruct (2509) Swiss AI38±4local~6.3GB2025-08
656LLaMA 65B Meta38±1comparison only2023-02
657Granite 3B Code Instruct (2K) IBM38±2local~3.7GB2024-04

see all LLMs · find the ones that fit your machine

$ what it's good at

math▁▂▃▅▆▅▄█▄▁▁▁ #367 of 398 * · 48% of its score · 2 tests

Beats Yi 34B Chat, Command R (03-2024), Molmo 7B-D (0924) here, though they rank higher overall. Trails Gemma 3 1B, OLMo 2 1B Instruct (0425), Stable Code 3B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
MATH (subset unspecified)21.9% ✎██····██·███ beats 0/6easy · avg 71%▇▇▇▇▇▇▇▇▇▇▇▇ 29%at its level · noisy
OMEGA (math generalisation)5.0% ✎█···▄·▄·▄··█ beats 0/6hard · avg 20%▇▇▇▇▇▇▇▇ 19%typical
other tests3 not in a task · 46% of its score
testresultvs the others in its rundifficultycountswhy
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy80.3% ✎▁▁▁▁▁▁▁▂▄█▄▃ beats 68/146easy · avg 83%▇▇▇▇▇▇▇ 18%at its level · noisy
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy58.1% ✎▃·▁▁▂▁▂▅█▇▄▁ beats 31/100medium · avg 68%▇▇▇▇▇▇▇ 17%at its level · noisy
Icelandic LLM Leaderboard average (Miðeind, Sept 2026 suite)7.7% ✎▃▁▃▂▂▄▁▃▄▄▄█ beats 0/38very hard · avg 14%▇▇▇▇ 11%too hard for it

$ 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 Apertus 8B Instruct (2509) · needs ~5.1 GB · from huggingface.co/unsloth
    chat in the terminal
    llama cli -hf unsloth/Apertus-8B-Instruct-2509-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf unsloth/Apertus-8B-Instruct-2509-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 Apertus 8B Instruct (2509) · needs ~4.53 GB, Mac only · from huggingface.co/mlx-community

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/mlx-community/Apertus-8B-Instruct-2509-4bit -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 Apertus 8B Instruct (2509) · needs ~5.1 GB · from huggingface.co/unsloth
    chat in the terminal
    ollama run hf.co/unsloth/Apertus-8B-Instruct-2509-GGUF:Q4_K_M

MLX (Mac)

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run Apertus 8B Instruct (2509) · needs ~4.53 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/Apertus-8B-Instruct-2509-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/Apertus-8B-Instruct-2509-4bit

Other ways to run it

  1. Run it as published · needs ~16.11 GB, NVIDIA only · from huggingface.co/swiss-ai (official)
    vllm serve swiss-ai/Apertus-8B-Instruct-2509

Official weights and documentation: huggingface.co/swiss-ai ↗

$ how the score was worked out · allenai/Olmo-3-7B-Instruct mo… 48%, EuroEval leaderboards (result… 36%, Miðeind Icelandic LLM Leaderb… 11%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
allenai/Olmo-3-7B-Instruct model card2▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 48%Apache-2.0
EuroEval leaderboards (results collection), English generative leaderboard2▇▇▇▇▇▇▇▇▇▇▇ 36%MIT
Miðeind Icelandic LLM Leaderboard1▇▇▇ 11%CC BY 4.0
prior · pull toward the middle▇▇ 6%
weighted average= 38.1

$ 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