PiCarX Connecting to Ollama online

General
1

When i add the base url as instructed to the code i get an error stating mulitple values.

from picarx.llm import OpenAI
from secret import OPENAI_API_KEY

INSTRUCTIONS = “You are a helpful assistant.”
WELCOME = “Hello, I am a helpful assistant. How can I help you?”

llm = OpenAI(
base_url=“https://api.openai.com/v1”,
api_key=OPENAI_API_KEY,
model=“gpt-4o”,
)

Set how many messages to keep

llm.set_max_messages(20)

Set instructions

llm.set_instructions(INSTRUCTIONS)

Set welcome message

llm.set_welcome(WELCOME)

print(WELCOME)

rk@pi:~/picar-x/example $ python3 18.online_llm_test.py
Traceback (most recent call last):
File “/home/rk/picar-x/example/18.online_llm_test.py”, line 7, in
llm = OpenAI(
base_url=“https://api.openai.com/v1”,
api_key=OPENAI_API_KEY,
model=“gpt-4o”,
)
File “/usr/local/lib/python3.13/dist-packages/sunfounder_voice_assistant/llm/init.py”, line 142, in init
super().init(*args,
~~~~~~~~~~~~~~~~^^^^^^^
base_url=“https://api.openai.com/v1”,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
**kwargs)
^^^^^^^^^
TypeError: sunfounder_voice_assistant.llm.llm.LLM.init() got multiple values for keyword argument ‘base_url’

2

AFAIK You should be using LLM not openai for ollama

from picarx.llm import LLM # not openai

Etc …

llm = LLM( # Not openai
    base_url="https://api.example.com/v1", # edit to whatever...
    api_key=API_KEY,
    model="your-model-name-here",          
)

The openai llm doesn’t have the base_url parameter.

3

Can you provide an example of what to use?

4

I’m not sure what more to add. The code i posted above is an example, just swap your equivalent lines for mine and fill in your providers url and your chosen model. I do not know what those are.

Or just use the final example directly from sunfounders site

Copied below….

from picarx.llm import LLM

from secret import API_KEY

INSTRUCTIONS = “You are a helpful assistant.”

WELCOME = “Hello, I am a helpful assistant. How can I help you?”

llm = LLM(

base_url="https://api.example.com/v1",  # fill in your provider’s base_url

api_key=API_KEY,

model="your-model-name-here",           # choose a model from your provider

)

5

I have tried using gemini and it works. But when i get to the last stage of adding a base url i get the same error stating the following.

TypeError: sunfounder_voice_assistant.llm.llm.LLM.init() got multiple values for keyword argument ‘base_url’

Adding the line base_url is causing an error with whatever LLM i choose. That command is obviously called somewhere else and is throwing there errors due to multiple calls.

6

Post your complete code for those llm models that you’ve tried. Include llm model chosen url and model. You say you’ve tried multiple ones, but just post a couple! Just to check for any theme.

I know there is another forum user here using ollama successfully on rpi. I’ve never used it myself, but ive had no problem with a couple of the other models ( although the api could have changed since I tried them) If the other user sees this chat, they may be able to post their code, otherwise Sunfounder will be more helpful than I when back in the office.

Sorry couldn’t really help.

7

I’m assuming it’s user error because the same result happens. Here is my code as i followed it according to the directions. Any help is greatly appreciated.

secret.py

Store secrets here. Never commit this file to Git.

GEMINI_API_KEY = “MY KEY HERE”

from picarx.llm import Gemini
from secret import GEMINI_API_KEY

INSTRUCTIONS = “You are a helpful assistant.”
WELCOME = “Hello, I am a helpful assistant. How can I help you?”

llm = Gemini(
base_url=“https://generativelanguage.googleapis.com”,
api_key=GEMINI_API_KEY,
model=“gemini-2.5-flash”,
)

rk@pi:~/picar-x/example $ sudo python3 18.online_llm_test.py
Traceback (most recent call last):
File “/home/rk/picar-x/example/18.online_llm_test.py”, line 7, in
llm = Gemini(
base_url=“https://generativelanguage.googleapis.com”,
api_key=GEMINI_API_KEY,
model=“gemini-2.5-flash”,
)
File “/usr/local/lib/python3.13/dist-packages/sunfounder_voice_assistant/llm/init.py”, line 213, in init
super().init(*args,
~~~~~~~~~~~~~~~~^^^^^^^
base_url=“https://generativelanguage.googleapis.com/v1beta/openai”,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
**kwargs)
^^^^^^^^^
TypeError: sunfounder_voice_assistant.llm.llm.LLM.init() got multiple values for keyword argument ‘base_url’

8

Check 1: Gemini API Key Validity

Ensure your key is obtained from Google AI Studio and is not expired or over its quota.

The key format should be: AIzaSyXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX (no extra spaces or characters).

You need to save the correctly formatted Gemini API key in the secret.py file.

Check 2: Base URL Validity
We support the OpenAI API format and any compatible APIs. Each API provider has its own API specification and base_url. Please refer to their documentation.
Since you are using the Gemini model, please verify the accuracy and correctness of the URL you have configured.

9

I did try OpenAI first. I have a key that works and it successfully ran the modules in the earlier sections. When I added the OpenAI base url as the instructions shown I would get that error message as well. I used https://api.openai.com/v1 as the address. The Gemini key is valid as well and the base url is correct. From my research it appears that base_url is being used somewhere else in the programming. I chose Gemini to confirm my earlier findings.

10

Here: https://github.com/sunfounder/sunfounder-voice-assistant/blob/main/sunfounder_voice_assistant/llm/init.py

The base_url has been pre-configured for you. Therefore, when you select libraries such as Ollama, OpenAI, or Gemini, you do not need to configure the base_url. You only need to configure the API Key and model. Please refer to this example:

If you do not want to use our pre-configured OpenAI setup, you can directly use the LLM class as follows:

from picarx.llm import LLM

llm = LLM(
base_url=“example.com/”,
api_key=OPENAI_API_KEY,
model=“gpt-4o”,
)

This allows you to specify your own base_url.