feat(opencode): add Open WebUI provider

[SamirMoustafa]

Issue for this PR

Closes #13537

Builds on feat: add provider openWebUI #14341 (thanks @PapeThePope).

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

Adds Open WebUI as a provider (OpenAI-compatible API): login/connection in CLI and TUI, listing and models, handling for gzip/chunked streams, docs, and tests.

How did you verify your code works?

• bun typecheck in packages/opencode
• Manual: Open WebUI URL + API key via CLI/TUI, model list and a short chat/streaming run

Screenshots / recordings

N/A (CLI/TUI text flows).

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

Hey! Your PR title Add Open WebUI Provider doesn't follow conventional commit format.

Please update it to start with one of:

• feat: or feat(scope): new feature
• fix: or fix(scope): bug fix
• docs: or docs(scope): documentation changes
• chore: or chore(scope): maintenance tasks
• refactor: or refactor(scope): code refactoring
• test: or test(scope): adding or updating tests

Where scope is the package name (e.g., app, desktop, opencode).

See CONTRIBUTING.md for details.

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

Potential Duplicate Found:

• PR feat: add provider openWebUI #14341: feat: add provider openWebUI
  • This PR is directly related as it's the original work that PR feat(opencode): add Open WebUI provider #18306 is built upon. The current PR (feat(opencode): add Open WebUI provider #18306) explicitly mentions it builds on PR feat: add provider openWebUI #14341.

The current PR (#18306) appears to be an updated or refactored version of the Open WebUI provider feature originally introduced in PR #14341. The description acknowledges this relationship, so these two PRs are related but the current one seems to be a successor/improvement to the earlier work.

[github-actions]

Thanks for updating your PR! It now meets our contributing guidelines. 👍

[SamirMoustafa]

Open WebUI + Ollama (local smoke test)

Download docker-compose.yml

docker compose up -d
docker compose exec ollama ollama pull qwen2.5:0.5b

Open http://localhost:3001 → admin signup → Settings → Account → API keys → create a key.

// Optional
export OPEN_WEBUI_BASE_URL=http://localhost:3001
export OPEN_WEBUI_API_KEY=…

Stop: docker compose down (volumes: docker compose down -v).
Ports: Web UI defaults to 3001 (OPEN_WEBUI_PORT in docker-compose.yml). Run docker compose only under infra/open-webui-local or add -f infra/open-webui-local/docker-compose.yml.

Screen.Recording.2026-03-20.at.2.15.07.AM.mov

[SamirMoustafa]

Hi @adamdotdevin and @alexyaroshuk,
Please review this PR when you have a chance.
It introduces the Open WebUI provider and updates several core provider files, CLI/TUI components, and related documentation.

Download docker-compose.yml

docker compose up -d
docker compose exec ollama ollama pull qwen2.5:0.5b

Besides the commented Dockerfile, I have tested over a version of Open WebUI with these providers (anthropic.claude-haiku-4-5, anthropic.claude-sonnet-4-6, anthropic.claude-opus-4-6, google.gemini-3-flash-preview, google.gemini-3.1-pro-preview, openai.gpt-5.4-nano, openai.gpt-5.4-mini, openai.gpt-5.4, azure.gpt-5-mini, azure.gpt-oss-120b, azure.DeepSeek-V3.2, azure.Kimi-K2.5, mistral.mistral-small-2603, mistral.codestral-2508)
I'm happy to review any comments, so please let me know.

[MikeWang0316tw]

Great work on this! Quick question — is there a way to configure contextLength and maxTokens for Open WebUI models? Since they're dynamically discovered and not in models.dev, I wasn't sure if there's a supported way to set these, or if it needs to be done manually in opencode.json.

[SamirMoustafa]

Great work on this! Quick question — is there a way to configure contextLength and maxTokens for Open WebUI models? Since they're dynamically discovered and not in models.dev, I wasn't sure if there's a supported way to set these, or if it needs to be done manually in opencode.json.

@MikeWang0316tw
Yes — for Open WebUI models (discovered at runtime), configure overrides manually in your project config.

Use either opencode.json, opencode.jsonc, or config.json at your project root.

{"$schema": "https://opencode.ai/config.json",
  "provider": {
    "openwebui": {
      "options": {
        "baseURL": "http://localhost:3001"
      },
      "models": {
        "qwen3.5:9b": {
          "limit": {
            "context": 8192,
            "output": 1024
          },
          "options": {
            "num_ctx": 8192,
            "num_predict": 512,
            "temperature": 0.2,
            "top_p": 0.9,
            "repeat_penalty": 1.05}}}}}}

In this example:

models.<id>.limit.context controls OpenCode context budgeting/compaction behavior.
models.<id>.limit.output drives maxOutputTokens in OpenCode.
models.<id>.options passes backend-specific request fields to your Open WebUI/Ollama stack.

A similar manual custom provider config flow in OpenCode is llama.cpp.

[richardclli]

When will this merge? I need this feature ^.^

[richardclli]

@SamirMoustafa You may need to resolve the conflict