hermes - 💡(How to fix) Fix WSL2 + Ollama on Windows: APIConnectionError on /v1 endpoint despite curl working [1 pull requests]

Environment:

• Hermes Agent latest
• WSL2 (Ubuntu) on Windows 11
• Ollama running on Windows (not in WSL)
• Model: gemma-4-abliterated with OLLAMA_CONTEXT_LENGTH=65536
• No GPU, CPU only

Problem: curl http://127.0.0.1:11434 → "Ollama is running" ✓ curl http://127.0.0.1:11434/v1/models → returns model list ✓ curl http://127.0.0.1:11434/v1/chat/completions → empty response instantly ✗ Hermes → APIConnectionError on every attempt ✗

Config: model: api_key: ollama base_url: http://127.0.0.1:11434/v1 context_length: 65536 default: gemma-4-abliterated model: gemma-4-abliterated provider: ollama-launch fallback_providers: []

What I tried:

• OLLAMA_HOST=0.0.0.0
• Hyper-V firewall rules
• port proxy via netsh
• localhostForwarding in wsl.conf
• Disabling Windscribe VPN

None worked. /v1/chat/completions returns empty body instantly from WSL even with model loaded at 65536 context.