Bronze
Back the work
$10 / month
- Your name on our supporters list
- A vote on what we build next
- A thank you in the build log
Open-weight model
Finance Chat
An open AI model for finance and money questions in plain chat. It runs fully offline on a small desktop, so account details and deal terms never leave your machine.
- Field
- Finance
- Runs
- Fully offline
- Built on
- AdaptLLM finance-chat
- License
- Free to use
Finance Chat
Finance Chat is an open AI model for money and finance questions in plain chat. It runs fully offline on a small desktop, so account numbers, deal terms, and other private figures never leave your machine.
What it can do
It talks through finance ideas in simple words: what a margin is, how a balance sheet fits together, or what a term in a deal means. It is built on AdaptLLM’s finance-chat, an open model trained on finance text, and packed into ready-to-run files for a single desktop.
How well it works
We scored five builds on FinanceBench, a strict 50-question test that only counts an exact number as right. The scores are low, 14 to 18 percent, and this is the honest weak spot: the model is good at explaining finance in words, but not at pulling exact figures out of a filing. So use it to learn and to draft, and check every number yourself. One nice find from our testing: the Q8_0 build matches the full-size model almost exactly while taking far less space.
How to run it
Download the GGUF files (the ready-to-run format) and run them with llama.cpp on a Spark-class desktop, a small AI machine with 128 GB of memory. The Q4_K_M build is the fastest, at about 31 tokens a second, and a good place to start.
Install
huggingface-cli download Orionfold/finance-chat-GGUFllama-cli -hf Orionfold/finance-chat-GGUF:Q4_K_MUse it
llama-cli -hf Orionfold/finance-chat-GGUF:Q4_K_M -p "Explain the difference between gross margin and net margin."from llama_cpp import Llama
llm = Llama.from_pretrained(
repo_id="Orionfold/finance-chat-GGUF",
filename="*Q4_K_M.gguf",
)
out = llm("Explain the difference between gross margin and net margin.")
print(out["choices"][0]["text"])
Specs
- Base model
- AdaptLLM/finance-chat
- Format
- GGUF (ready to run)
- Builds
- Q4_K_M · Q5_K_M · Q6_K · Q8_0 · F16
- Best build
- Q4_K_M (about 31 tokens a second on a Spark desktop)
- License
- Free to use
Benchmarks
| Build | FinanceBench score | Speed on a Spark |
|---|---|---|
| Q4_K_M (fastest) | 14% | 31 tokens a second |
| Q5_K_M | 16% | 27 tokens a second |
| Q6_K | 16% | 24 tokens a second |
| Q8_0 | 18% | 9 tokens a second |
| F16 (full size) | 18% | 12 tokens a second |
Used in the open
Live counts from HuggingFace, refreshed when the site builds. Built and maintained in the open by Orionfold.
- 531
- Downloads · last 30 days
Sponsor Finance Chat
Back this work with a monthly tier. Your support moves your requests up the list, and Gold or Platinum earns a badge on the roadmap item you back.
Silver
Get a say
$25 / month
- Everything in Bronze, plus:
- Your requests join the priority line
- Early access to new versions
Gold
Move it up the list
$50 / month
- Everything in Silver, plus:
- High priority on your feature and fix requests
- A Gold badge on the roadmap item you back
- A monthly call to plan the work
Platinum
Shape the roadmap
$100 / month
- Everything in Gold, plus:
- Top of the line, we work your requests first
- A Platinum badge on the roadmap item you back
- A direct line to us and a hand in the roadmap
Need something specific? Send an enquiry from the roadmap.
Keep exploring
Model
Patent Strategist
Offline patent reasoning in ready-to-run files, built with the NeMo toolkit. Nothing leaves your desktop.
Book
AI Research on NVIDIA DGX Spark
Real notes from doing AI research on one desktop. The NVIDIA DGX Spark is a small machine with huge power (petascale means it runs about a quadrillion math steps a second), so you can push local AI further with no cloud needed. Every lesson is backed by code that runs.