whatmodel.sh / watt-tool-70B

watt-tool-70B

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

watt-tool-70B is an open-weights language model from Watt AI, released December 2024. Its published size is 70.55B parameters. Licence: apache-2.0.

Catalogue identity notes

Maker repo watt-ai/watt-tool-70B @dbe1934: licence tag apache-2.0. 70,553,706,496 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2024-12: month the maker's Hugging Face repo was created (createdAt 2024-12-19).

66.7±6.8#123 of 949 LLMsfrom 1 benchmark result across 1 source* thinly tested, treat the rank as provisional

$ how it compares

The LLMs ranked either side of watt-tool-70B, out of 1132 with a score, then more from Watt AI.

#modelscorewhereRAMreleased
155Gemini 2.5 Pro Preview (03-25) Google67±2comparison only2025-04
156Gemini 2.5 Pro Preview (05-06) Google67±1comparison only2025-05
157watt-tool-70B Watt AI67±7local~22GB2024-12
158GigaChat 3.5 Ultra Reasoning (432B-A28B) Sber67±1local~461GB2026-09
159Qwen3.5 35B-A3B Alibaba67±0api $0.35/M +local~23GB2026-02
more from Watt AI
249watt-tool-8B Watt AI62±7local~6.1GB2024-12

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▄▄▃▂▁ #151 of 334 * · 79% of its score · 1 test

Trails SmolLM3 3B, Jamba 1.5 Mini, Jamba 1.5 Large here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
BFCL v373.6% ✎▁▁▁▁▁▃▅█▅▄▃▁ beats 112/114medium · avg 52%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 79%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 watt-tool-70B · needs ~42.5 GB · from huggingface.co/bartowski
    chat in the terminal
    llama cli -hf bartowski/watt-ai_watt-tool-70B-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf bartowski/watt-ai_watt-tool-70B-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 watt-tool-70B · needs ~42.5 GB · from huggingface.co/bartowski

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/bartowski/watt-ai_watt-tool-70B-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 watt-tool-70B · needs ~42.5 GB · from huggingface.co/bartowski
    chat in the terminal
    ollama run hf.co/bartowski/watt-ai_watt-tool-70B-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 watt-tool-70B · needs ~75 GB, Mac only · from huggingface.co/hellosri
    chat in the terminal
    mlx_lm.chat --model hellosri/watt-tool-70B-mlx-8Bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model hellosri/watt-tool-70B-mlx-8Bit

Other ways to run it

  1. Run it as published · needs ~141.11 GB · from huggingface.co/watt-ai (official)
    uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='watt-ai/watt-tool-70B', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"

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

$ how the score was worked out · Berkeley Function Calling Lea… 79%

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 archive1▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 79%Apache-2.0
prior · pull toward the middle▇▇▇▇▇▇ 21%
weighted average= 66.7

$ 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