[Bug]: Pythagora cannot start because the LLM API is not reachable.
Author: blapCreated Aug 8, 2024Updated Sep 15, 2026
Labelsbug
Version
Command-line (Python) version
Operating System
Windows 10
What happened?
Edited config.json to use OpenRouter:
// Configuration for the LLM providers that can be used. Pythagora supports
// OpenAI, Azure, Anthropic and Groq. OpenRouter and local LLMs (such as LM-Studio)
// also work, you can use "openai" provider to define these.
"llm": {
"openai": {
// Base url to the provider/server, omitting the trailing "chat/completions" part.
"base_url": "https://openrouter.ai/api/v1",
"api_key": "key",
"connect_timeout": 60.0,
"read_timeout": 10.0
},
// Each agent can use a different model or configuration. The default, as before, is GPT4 Turbo
// for most tasks and GPT3.5 Turbo to generate file descriptions. The agent name here should match
// the Python class name.
"agent": {
"default": {
"provider": "openai",
"model": "recursal/eagle-7b"
running python main.py:
[Pythagora] Pythagora cannot start because the LLM API is not reachable.
What I did wrong?
Source: Pythagora-io/gpt-pilot