status_code: 404, model_name: deepseek-chat

[sliverp]

Question

I set .env file as follows:

but got

pydantic_ai.exceptions.ModelHTTPError: status_code: 404, model_name: deepseek-chat, body:
An unexpected error occurred: status_code: 404, model_name: deepseek-chat, body:
Ask a question (Press Ctrl+J to submit, Enter for new line):

what shoud i do?

Context

Any relevant context or background information

What I've Tried

I print the AppConfig and find my config is loaded successfully

[vitali87]

@sliverp The issue is that openai provider uses the Responses API (/responses endpoint), but DeepSeek only supports the Chat Completions API (/chat/completions).

Workaround: Use ollama as the provider - it uses the compatible Chat Completions API:

ORCHESTRATOR_PROVIDER=ollama
ORCHESTRATOR_MODEL=deepseek-chat
ORCHESTRATOR_ENDPOINT=https://api.deepseek.com/v1
ORCHESTRATOR_API_KEY=your-key

CYPHER_PROVIDER=ollama
CYPHER_MODEL=deepseek-chat
CYPHER_ENDPOINT=https://api.deepseek.com/v1
CYPHER_API_KEY=your-key

[vitali87]

@sliverp let me know if it worked for you, and we will close this issue.

[AoXiang-Soar]

@sliverp let me know if it worked for you, and we will close this issue.

Using Ollama as the provider can indeed resolve this issue, but it also introduces another problem. The reason lies in the check_ollama_running function in codebase_rag/providers/base.py—this function attempts to construct a URL ending with /api/tags based on the original URL and requests a response. However, most LLM providers do not implement this functionality, causing the request to fail and ultimately leading to startup failure. In fact, this is exactly how I discovered the issue #284. To support more LLM providers, I recommend re-implementing a custom provider.

[evander-wang]

Question

I set .env file as follows:

but got

pydantic_ai.exceptions.ModelHTTPError: status_code: 404, model_name: deepseek-chat, body: An unexpected error occurred: status_code: 404, model_name: deepseek-chat, body: Ask a question (Press Ctrl+J to submit, Enter for new line):

what shoud i do?

Context

Any relevant context or background information

What I've Tried

I print the AppConfig and find my config is loaded successfully

Edit the file providers/base.py, to tmp fix and maintain your deepseek-cat config unchanged