whatmodel.sh / ThinkAgent-1B (Llama-3.2-1B-FC-v3)

ThinkAgent-1B (Llama-3.2-1B-FC-v3)

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

ThinkAgent-1B (Llama-3.2-1B-FC-v3) is an open-weights language model from ThinkAgents, released January 2025. Its published size is 1.24B parameters. Licence: apache-2.0.

Catalogue identity notes

Maker repo AymanTarig/Llama-3.2-1B-FC-v3 @5fad827 (requested as ThinkAgents/ThinkAgent-1B, which the HF API redirects to AymanTarig/Llama-3.2-1B-FC-v3): licence tag apache-2.0. 1,235,814,400 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2025-01: month the maker's Hugging Face repo was created (createdAt 2025-01-31).

25.5±7.0#892 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 ThinkAgent-1B (Llama-3.2-1B-FC-v3), out of 1132 with a score.

#modelscorewhereRAMreleased
1069GPT-3 curie (API base model) OpenAI26±1comparison only2020-06
1070Alpaca 13B Stanford CRFM26±2comparison only2023-03
1071ThinkAgent-1B (Llama-3.2-1B-FC-v3) ThinkAgents26±7local~4.1GB2025-01
1072Falcon 7B Instruct TII25±2local~5.9GB2023-04
1073Qwen2 0.5B Instruct Alibaba25±2local~1.8GB2024-06

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▃▁▃▆█▆▅▄▄▃▂▁ #331 of 334 * · 78% of its score · 1 test

Beats QwQ-32B-Preview, Gemma 3 1B here, though they rank higher overall.

testresultvs the others in its rundifficultycountswhy
BFCL v320.8% ✎▁▃▁▁▁▃▅█▅▄▁▁ beats 3/114medium · avg 52%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 78%blunt · at its level

$ where to run it

Other ways to run it

  1. Run it as published · needs ~2.47 GB · from huggingface.co/AymanTarig
    uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='AymanTarig/Llama-3.2-1B-FC-v3', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"

Weights and documentation: huggingface.co/AymanTarig ↗

$ how the score was worked out · Berkeley Function Calling Lea… 78%

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▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 78%Apache-2.0
prior · pull toward the middle▇▇▇▇▇▇▇ 22%
weighted average= 25.5

$ 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