ollama - 💡(How to fix) Fix ollama launch opencode generates invalid OpenCode config when local models have context length but no max output tokens [1 pull requests]

What is the issue?

Ollama version: 0.30.0 OpenCode version: 1.15.13 macOS

Steps:

1. Have local Ollama models, e.g. qwen3.6:35b-a3b
2. Run: ollama launch opencode
3. Select the model

Expected: OpenCode TUI starts.

Actual: OpenCode fails during startup: Error: 4 of 5 requests failed Affected startup requests: config.providers, provider.list, app.agents, config.get

OpenCode log shows: ConfigInvalidError / SchemaError: Missing key at ["provider"]["ollama"]["models"][...]["limit"]["output"]

Root cause: cmd/launch/opencode.go buildModelEntries() emits: "limit": { "context": 262144 } when model.ContextLength > 0 but model.MaxOutputTokens == 0.

OpenCode expects limit.output when limit is present.

Suggested fix: When generating OpenCode model entries, either:

• include both limit.context and limit.output, using a sane default for output when MaxOutputTokens is unknown, or
• omit limit entirely unless both fields are known, or
• fetch/derive max output tokens for local models.

Relevant log output

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.30.0