whatmodel.sh / Nanbeige4-3B-Thinking-2511

Nanbeige4-3B-Thinking-2511

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

Nanbeige4-3B-Thinking-2511 is an open-weights language model from Nanbeige, released November 2025. Its published size is 3.93B parameters. Licence: apache-2.0.

Catalogue identity notes

Maker repo Nanbeige/Nanbeige4-3B-Thinking-2511 @396a1fb: licence tag apache-2.0. 3,933,637,120 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2025-11: month the maker's Hugging Face repo was created (createdAt 2025-11-20).

54.2±3.1#345 of 949 LLMsfrom 2 benchmark results across 2 sources* thinly tested, treat the rank as provisional

$ how it compares

The LLMs ranked either side of Nanbeige4-3B-Thinking-2511, out of 1132 with a score.

#modelscorewhereRAMreleased
448Devstral Medium 1.0 (2507) Mistral54±3comparison only2025-07
449Mercury Inception54±3comparison only
450Nanbeige4-3B-Thinking-2511 Nanbeige54±3local~3.7GB2025-11
451Mellum2 12B-A2.5B Thinking JetBrains54±1local~8.6GB2026-05
452Mistral Large 2 (2407) Mistral54±1local~45GB2024-07

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▄▄▃▂▁ #148 of 334 * · 42% of its score · 1 test

Beats watt-tool-70B, Qwen3 235B-A22B Thinking 2507, o4-mini here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Jamba 1.5 Large here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
BFCL V4 (December 2025 archive)51.4% ✎▁▁▁▂██▅▅▆▂▂▂ beats 58/77hard · avg 32%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 42%at its level
other tests1 not in a task · 54% of its score
testresultvs the others in its rundifficultycountswhy
EQ-Bench 3 (emotional intelligence, judge Elo)1014.3 ✎▁·▂▃▄▅▅▃█▄▁▁ beats 4/62avg 1044▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 54%typical

$ 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 Nanbeige4-3B-Thinking-2511 · needs ~2.4 GB · from huggingface.co/bartowski
    chat in the terminal
    llama cli -hf bartowski/Nanbeige_Nanbeige4-3B-Thinking-2511-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf bartowski/Nanbeige_Nanbeige4-3B-Thinking-2511-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 Nanbeige4-3B-Thinking-2511 · needs ~2.4 GB · from huggingface.co/bartowski

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/bartowski/Nanbeige_Nanbeige4-3B-Thinking-2511-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 Nanbeige4-3B-Thinking-2511 · needs ~2.4 GB · from huggingface.co/bartowski
    chat in the terminal
    ollama run hf.co/bartowski/Nanbeige_Nanbeige4-3B-Thinking-2511-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 Nanbeige4-3B-Thinking-2511 · needs ~2.1 GB, Mac only · from huggingface.co/RepublicOfKorokke
    chat in the terminal
    mlx_lm.chat --model RepublicOfKorokke/Nanbeige4-3B-Thinking-2511-mlx-mxfp4
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model RepublicOfKorokke/Nanbeige4-3B-Thinking-2511-mlx-mxfp4

Other ways to run it

  1. Run it as published · needs ~7.87 GB · from huggingface.co/Nanbeige (official)
    uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='Nanbeige/Nanbeige4-3B-Thinking-2511', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"

Official weights and documentation: huggingface.co/Nanbeige ↗

$ how the score was worked out · EQ-Bench 3 canonical leaderbo… 54%, Berkeley Function Calling Lea… 42%

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
EQ-Bench 3 canonical leaderboard results1▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 54%MIT
Berkeley Function Calling Leaderboard result archive1▇▇▇▇▇▇▇▇▇▇▇▇ 42%Apache-2.0
prior · pull toward the middle▇ 4%
weighted average= 54.2

$ 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