You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
BREAKING CHANGES:
- MODEL_PROVIDERS default changed from None to 'ALL'
- pipe() is now async with __user__ parameter (Open WebUI v0.4+)
Features:
- Pre-flight API key validation via /auth/key
- Provider routing (sort, prefer, exclude, require_parameters, data_collection)
- Model fallbacks with automatic failover
- Middle-out compression for long prompts
- Anthropic-style cache_control injection
- Configurable reasoning effort (low/medium/high) with <think> tags
- 22 provider icons for visual model identification
- Citation injection with markdown links
- Auto-retry on timeout/connection errors (configurable MAX_RETRIES)
- FREE_ONLY now detects free models by pricing (not just :free suffix)
Bug fixes:
- Stream exception handlers now close </think> tag if open
- HTTPError in stream: body cached before response.close()
- Empty choices array no longer causes IndexError
- response=None in _format_http_error handled gracefully
- Open WebUI internal keys stripped from payload
- Dict user field no longer forwarded to OpenRouter
Infrastructure:
- GitHub Actions CI (Python 3.10-3.13)
- 170 unit tests + 47 integration tests
- Issue templates, security policy, sponsor config
- Pre-release testing checklist (TESTING.md)
- Pydantic v2 compatibility (removed from __future__ import annotations)
OpenRouter Pipe is the most feature-complete integration between [Open WebUI](https://docs.openwebui.com) and [OpenRouter](https://openrouter.ai). It gives you access to **300+ AI models** — including GPT-5, Claude 4, Gemini 2.5, Llama 4, DeepSeek R1, and more — directly in your Open WebUI interface, with zero configuration beyond an API key.
27
+
28
+
**Key differentiators:**
29
+
-**Pre-flight API key validation** — invalid keys are caught before you see any models, not after you send a message
30
+
-**Full provider routing** — sort, prefer, exclude, and require parameters across providers
31
+
-**Native reasoning tokens** — `<think>` blocks with configurable effort levels
0 commit comments