Skip to content

desktop app: restore a way to see current model details (cost, variants, limits) in the GUI #59818

Open

Description

@pkjoe88

Summary

The desktop app used to show a persistent right-side panel with the details of the currently selected model (Provider, model ID, variants, cost per Mtok, context/output limits, capabilities). In the current desktop app (2.2), this panel is gone and I cannot find any way to bring it back — the model picker no longer shows cost, variants, or detailed limits either. Requesting a way to see current-model details in the GUI again (a collapsible sidebar, or details in the model picker).

Environment

  • opencode version: 2.2.0 (desktop app)
  • OS: Windows_NT 10.0.26100 (win32 x64)
  • Terminal: Unavailable: terminal environment variables are not set (issue is in the desktop app GUI)
  • Shell: C:\Windows\system32\cmd.exe (only used for CLI checks)
  • Install/channel: latest
  • Active plugins: opencode-antigravity-auth@latest

Reproduction

  1. Open a session in the desktop app.
  2. Open the model picker (click the model chip near the composer).
  3. Look for cost per token, variants (reasoning effort: low/medium/high), and detailed context/output limits for the selected model.

Expected Behavior

Either:

  1. A right-side info pane shows the current model's Provider, model ID, variants, cost per Mtok, context/output limits, and capabilities (ideally collapsible, remembering its state), or
  2. The model picker shows cost / variants / context limit in its rows or in a details popover when a model is selected.

Actual Behavior

No persistent or on-demand surface in the GUI shows this information. The picker displays only names and simple labels; cost, variants, cache pricing, and detailed limits are not visible anywhere in the app.

Additional Context

Why this matters:

  • Verify at a glance which variant (e.g. reasoning effort: low/medium/high) the session is actually using.
  • Check cost per input/output/cache token before switching to a paid model, without switching to a terminal.
  • See context/output limits to anticipate compaction before a long session.
  • Free models ("$0") vs paid models are easy to confuse when costs are not shown.

What exists today: opencode models --verbose in the CLI prints this information, but the desktop app has no equivalent surface. Even option 2 above would fully cover the use case.

Happy to provide screenshots or reproduction details if useful.

Activity

  1. opencodejr commented on Oct 28, 2026

    @opencodejr
    Contributor

    Thanks for the detailed request! You're asking for a way to see the current model's details in the desktop app: provider, model ID, variant, cost per input/output/cache token, and context/output limits. That could be a collapsible side panel or a details view in the model picker, so you don't have to run opencode models --verbose in a terminal.

    For context: the hover card on the composer's model chip showed provider, context limit, modalities, tools and reasoning. It was removed on purpose in #59108, but that change isn't in a release yet, so 2.2.0 should still show the card when you hover the chip. The same card appears when you hover a model inside the experimental new model picker. Neither has ever shown cost, cache pricing or output limits, though, so what you're asking for would be new.

    Maintainers will consider it. Related issues:

  2. pkjoe88 commented on Oct 28, 2026

    @pkjoe88
    Author

    Thanks @opencodejr[bot] for the triage and the related-issue links.

    Two additions for anyone considering adoption:

    1. Linking the removal: Since the hover card on the composer's model chip is being intentionally removed in fix(app): drop model details tooltip from the composer model chip #59108 (not yet shipped), the removal and this request should be evaluated together rather than separately. If that PR lands without an equivalent surface, the desktop app will have no way to see cost, cache pricing, or output limits at all — not even the context-limit line the card used to show for a lot of models.

    2. Community signal: If you've run into this — for example, switching models without knowing the cost, or not knowing whether a model has reasoning variants — please react with 👍 on this issue (and on [Desktop/Web] Model selector should scale to the available screen size (fixed ~284px popover feels cramped) #48789 / desktop: custom provider form cannot set context/output limits, reasoning effort, or discover models #57325, which are related surface limitations). Reactions are what make feature requests visible to maintainers.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions