[Bug]: Unable to use locally deployed Ollama models

[FUA-Studio]

Affected Component

External Integrations (LLM/Search APIs)

Describe the bug

PentAGI fails to start and cannot use locally deployed Ollama models, due to a hardcoded custom provider validation error requiring an OpenAI API key.（maybe）
I have modified the docker-compose configuration and the .env environment file.
Helpppppp!!!!!!!!

Steps to Reproduce

1. Pull the official vxcontrol/pentagi docker image (version 2.0.0 / 2.1.0)
2. Create docker-compose.yml and .env files, set environment variables including OPENAI_API_KEY with a dummy key and Ollama local server config
3. Run docker compose up -d to start the PentAGI service
4. Observe the container logs: the service crashes repeatedly with the error message: "LLM provider controller initialization failed: failed to create custom provider: missing the OpenAI API key, set it in the OPENAI_API_KEY environment variable"
5. Even after correctly setting OPENAI_API_KEY environment variable and modifying all *.provider.yml config files, the error persists, preventing the service from starting normally and connecting to local Ollama models

System Configuration

PentAGI version: v2.1.0-879e87c (official vxcontrol/pentagi docker image), also verified on v2.0.0
Deployment type:

• Docker Compose

Environment:

• Docker version: 27.x.x
• Docker Compose version: v2.x.x
• Host OS: Windows 10 / Ubuntu 22.04
• Available resources:
  • Memory: 16GB
  • CPU: 8 cores
• Local Ollama service running on host port 11434 (OpenAI compatible API)

Enabled features:

• Langfuse analytics: disabled
• Grafana monitoring: disabled
• Custom .env config: OPENAI_API_KEY set with dummy value, Ollama server URL configured

Logs and Artifacts

INFO[0000] Starting PentAGI 2.1.0-879e87c
DEBU[0000] Debug logging enabled
2026/07/17 06:02:09 goose: no migrations to run. current version: 2.1.0-879e87c
2026/07/17 06:02:09 Database schema updated successfully
INFO[0000] start profiling server on :7777 component=init
DEBU[0000] Docker client initialized client_version=... data_dir=/opt/pentagi/data docker_api_version=... docker_server_version=... host_dir=...
2026/07/17 06:02:09 LLM provider controller initialization failed: failed to create custom provider: missing the OpenAI API key, set it in the OPENAI_API_KEY environment variable

Screenshots or Recordings

none

Verification

• I have checked that this issue hasn't been already reported
• I have provided all relevant configuration files (with sensitive data removed)
• I have included relevant logs and error messages
• I am running the latest version of PentAGI

[asdek]

hey @FUA-Studio

you are using wrong variable name OPENAI_API_KEY, it should be OPEN_AI_KEY

please look at here, you should use this variable only with original OpenAI provider or reverse proxy to OpenAI provider combined with OPEN_AI_SERVER_URL, for any OpenAI-compatible providers please use custom provider.

Ollama supported as well, details here

[FUA-Studio]

我需要完整的 .env 以及 Docker Compose的文件内容 大模型为qwen2.5:14b

[FUA-Studio]

Require full .env and docker-compose.yml configurations, using qwen2.5:14b as the LLM.