How can I make the env file for the third party provider LLM model

[OVERSKY2003]

How can I make env file using the third party provider LLM model info as below.
base_url: https://api.cursorai.art/v1
chat model: gpt-5-2025-08-07
embedding model: text-embedding-3-small

the result of running "rdagent health_check" is ok

• 🧪 Testing embedding model: text-embedding-3-small
• ✅ Embedding test passed.
• 🧪 Testing chat model: gpt-5-2025-08-07
• ✅ Chat test passed.
• ✅ All tests completed

But the result of running "rdagent fin_quant" is Failed
RuntimeError: Failed to create chat completion after 10 retries.

[utsab345]

When integrating third-party LLM APIs, you need to tell RD-Agent where to find the model, what type it is, and how to authenticate.

Here’s an example .env file you can use:

# ===============================
# RD-Agent Third-Party LLM Config
# ===============================

# Third-party LLM provider base URL
LLM_API_BASE=https://api.cursorai.art/v1

# Chat model name (as provided by the third-party)
LLM_MODEL_NAME=gpt-5-2025-08-07

# Embedding model name
LLM_EMBEDDING_MODEL=text-embedding-3-small

# Your API key from the third-party provider
LLM_API_KEY=your_api_key_here

# (Optional) Timeout or retry configurations
LLM_REQUEST_TIMEOUT=60
LLM_MAX_RETRIES=5

# RD-Agent specific configurations
RDAGENT_MODE=production
RDAGENT_LOG_LEVEL=info

[SunsetWolf]

Hi, @OVERSKY2003

Based on the information provided so far, it’s difficult to determine the exact cause of the issue.

If you could share more details — for example, the full error message or traceback — it would help us better analyze and resolve the problem.

[OVERSKY2003]

@SunsetWolf , would you please give me the sample env file using the third party provider LLM model info? Your env configuration guide in github only demostrate the CHAT_MODEL config for openai, AZURE,deepseek.

[SunsetWolf]

@OVERSKY2003 , Since RD-Agent is built on LiteLLM, you can refer to the LiteLLM documentation for guidance on configuring other third-party providers. Hope this helps.