feat: discover Modal models

[devennavani]

Issue for this PR

Closes #14389.

Related to #6231.

Type of change

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

What does this PR do?

Modal model IDs are shared-endpoint hostnames scoped to the authenticated workspace, so they cannot be listed statically in models.dev.

This adds a built-in Modal provider plugin using OpenCode's existing provider.models hook. It fetches authenticated GET /v1/models, exposes the returned endpoint hostnames as callable model IDs, and consumes the returned model metadata, including reasoning variants and interleaved reasoning. Missing metadata falls back to the corresponding models.dev template. Discovery failures return no models rather than exposing non-callable template IDs.

The provider registration and Inkling fallback template landed in models.dev #3760.

How did you verify your code works?

• bun test test/plugin/modal-models.test.ts
• bun typecheck from packages/opencode
• pre-push bun turbo typecheck (30 packages)

Screenshots / recordings

Not applicable.

Checklist

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

[github-actions]

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

Potential duplicate PRs found:

1. PR feat(opencode): discover abacus models from /v1/models endpoint #34563: feat(opencode): discover abacus models from /v1/models endpoint

   • feat(opencode): discover abacus models from /v1/models endpoint #34563
   • Why related: This PR implements the same pattern of discovering models from a /v1/models endpoint with bearer authentication. The Abacus implementation appears to be a similar provider discovery mechanism that could inform or overlap with the Modal models approach.

2. PR feat(opencode): auto-discover models from OpenAI-compatible providers #32731: feat(opencode): auto-discover models from OpenAI-compatible providers

   • feat(opencode): auto-discover models from OpenAI-compatible providers #32731
   • Why related: Addresses model discovery from provider endpoints generically. The current PR mentions running "discovery loaders generically instead of special-casing GitLab," which suggests overlap with how this PR handles generic provider discovery.

[github-actions]

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