whatmodel.sh / FireFunction v1
FireFunction v1
↗ share
+ contribute
FireFunction v1 is an open-weights language model from Fireworks AI, released February 2024. Its published size is 46.7B parameters (12.9B active per token). Licence: apache-2.0.
Catalogue identity notes
Maker repo fireworks-ai/firefunction-v1 @d193253: licence tag apache-2.0. 46,702,809,088 parameters from the Hugging Face safetensors metadata; active 12.9B per token (Mixtral 8x7B architecture (MixtralForCausalLM, 8 experts, 2 per token); Mistral: 12.9B parameters per token). Released 2024-02: month the maker's Hugging Face repo was created (createdAt 2024-02-16).
$ how it compares
The LLMs ranked either side of FireFunction v1, out of 1132 with a score, then more from Fireworks AI.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 839 |  Ministral 3B (2410) | 42±1 | comparison only | ||
| 840 |  OLMo 2 7B Instruct (1124) | 42±1 | local | ||
| 841 | FireFunction v1 | 42±6 | local | ||
| 842 |  WizardCoder-15B-V1.0 | 42±1 | local | ||
| 843 |  Snowflake Arctic Instruct | 41±2 | local | ||
| more from Fireworks AI | |||||
| 568 | FireFunction v2 (70B) | 50±6 | local | ||
$ what it's good at
agent tasks▁▁▃▆█▆▅▄▄▃▂▁ #314 of 334 * · 81% of its score · 1 test
Beats GLM-4 9B Chat, DBRX Instruct, Olmo 3 7B Instruct here, though they rank higher overall. Trails Granite 4.0 H 350M, Qwen3 0.6B, Qwen2.5 1.5B Instruct here, though they rank lower overall.
$ where to run it
Other ways to run it
- Run it as published · needs ~93.41 GB · from huggingface.co/fireworks-ai (official)
uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='fireworks-ai/firefunction-v1', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"
Official weights and documentation: huggingface.co/fireworks-ai ↗
$ how the score was worked out · Berkeley Function Calling Lea… 81%
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
| source | results | share of the score |
|---|---|---|
 Berkeley Function Calling Leaderboard result archive | 1 | 81% |
| prior · pull toward the middle | 19% | |
| weighted average | = 41.5 |
$ improve this page
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