whatmodel.sh / Granite 20B FunctionCalling

Granite 20B FunctionCalling

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

Granite 20B FunctionCalling is an open-weights language model from IBM, released July 2024. Its published size is 20.07B parameters. Licence: apache-2.0.

Catalogue identity notes

Maker repo ibm-granite/granite-20b-functioncalling @b030a7f: licence tag apache-2.0. 20,066,993,152 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2024-07: month the maker's Hugging Face repo was created (createdAt 2024-07-09).

45.0±4.1#587 of 949 LLMsfrom 2 benchmark results across 1 source* thinly tested, treat the rank as provisional

$ how it compares

The LLMs ranked either side of Granite 20B FunctionCalling, out of 1132 with a score, then more from IBM.

#modelscorewhereRAMreleased
737QwQ-32B-Preview Alibaba45±2local~21GB2024-11
738DeepSeek-Coder-V2 Lite Instruct (16B-A2.4B) DeepSeek45±1local~11GB2024-06
739Granite 20B FunctionCalling IBM45±4local~15GB2024-07
740GPT-4o Audio Preview (2024-10-01) OpenAI45±3comparison only2024-10
741Claude 2.1 Anthropic45±1comparison only2023-11
more from IBM
227Granite 4.2 30B IBM63±1api $0.28/M +local~19GB2026-08
293Granite 4.2 8B IBM60±1api $0.11/M +local~6.6GB2026-08
356Granite 4.2 3B IBM57±1api $0.052/M +local~3.7GB2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▄▄▃▂▁ #289 of 334 * · 92% of its score · 2 tests

Beats DeepSeek R1 0528 Qwen3 8B, Magistral Small 1.0 (2506), Llama 4 Scout (109B-A17B) here, though they rank higher overall. Trails Granite 4.0 H 1B, Granite 4.0 1B, Qwen2.5 3B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
BFCL V4 (December 2025 archive)23.2% ✎▁▁▁▂██▅▅▆▂▂▂ beats 11/77hard · avg 32%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 58%at its level
BFCL v348.9% ✎▁▁▁▁▁▃▅█▅▄▁▁ beats 32/114medium · avg 52%▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 34%blunt · at its level

$ 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 Granite 20B FunctionCalling · needs ~12.8 GB · from huggingface.co/mradermacher
    chat in the terminal
    llama cli -hf mradermacher/granite-20b-functioncalling-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf mradermacher/granite-20b-functioncalling-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 Granite 20B FunctionCalling · needs ~12.8 GB · from huggingface.co/mradermacher

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/mradermacher/granite-20b-functioncalling-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 Granite 20B FunctionCalling · needs ~12.8 GB · from huggingface.co/mradermacher
    chat in the terminal
    ollama run hf.co/mradermacher/granite-20b-functioncalling-GGUF:Q4_K_M

Other ways to run it

  1. Run it as published · needs ~40.13 GB · from huggingface.co/ibm-granite
    uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='ibm-granite/granite-20b-functioncalling', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"

Weights and documentation: huggingface.co/ibm-granite ↗

$ how the score was worked out · Berkeley Function Calling Lea… 92%

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
Berkeley Function Calling Leaderboard result archive2▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 92%Apache-2.0
prior · pull toward the middle▇▇ 8%
weighted average= 45.0

$ 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