fix(providers): update xAI catalog default to grok-4.5

[lfaoro]

Add grok-4.5 as the curated xAI default and keep grok-4 as the previous flagship for offline/fallback model selection.

Summary by CodeRabbit

• Updates
  • Added xAI’s Grok 4.5 to the available model list.
  • Set Grok 4.5 as the default xAI model for new sessions and OAuth setup.
  • Kept Grok 4 available, labeled as the previous flagship model.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: f4833eb3-75a3-450f-82da-65046dfd13ed

📥 Commits

Reviewing files that changed from the base of the PR and between 5f08a1f and fe921a4.

📒 Files selected for processing (1)

• internal/tui/onboarding_test.go

---

Walkthrough

The xAI provider default changes from grok-4 to grok-4.5. The curated catalog adds grok-4.5, relabels grok-4 as the previous flagship, and updates OAuth onboarding test setup.

Changes

xAI model catalog

Layer / File(s)	Summary
Promote grok-4.5 as the default internal/providercatalog/catalog.go, internal/providermodelcatalog/catalog.go, internal/tui/onboarding_test.go	The provider default, curated model list, and OAuth test configuration now identify grok-4.5 as the catalog default, while grok-4 is labeled the previous flagship.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Possibly related PRs

• Gitlawb/zero#141: Introduces or structures the provider catalog definitions updated by this change.

Suggested reviewers: anandh8x, gnanam1990

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the main change: updating the xAI catalog default model to grok-4.5.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@internal/providercatalog/catalog.go`:
- Line 142: Update the onboarding test fixture in the relevant test setup to use
DefaultModel "grok-4.5" instead of "grok-4", and adjust its related assertions
to expect the promoted default while preserving the existing onboarding
behavior.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: c8580d98-346e-4e70-8d0e-92f8767b825d

📥 Commits

Reviewing files that changed from the base of the PR and between 6fc1220 and 5f08a1f.

📒 Files selected for processing (2)

• internal/providercatalog/catalog.go
• internal/providermodelcatalog/catalog.go

[jatmn]

I found issues that need to be addressed before this is ready.

Findings

• [P2] Register Grok 4.5's reasoning capability with the promoted default
  internal/providermodelcatalog/catalog.go:146
  The new default is only a bare curated model row. Because it has no model-registry entry and the unknown-model fallback has no Grok rule, the TUI's /effort command reports no controls for grok-4.5; users cannot select its documented low, medium, or high reasoning levels and are stuck with the provider default (high). Add its supported reasoning metadata (and regression coverage) to the capability path when promoting it as the default.

[gnanam1990]

Local review (not a blocker): built and ran touched-package tests locally; all pass. One verification ask before merge.

[gnanam1990]

[P2] Confirm grok-4.5 is a live model ID on api.x.ai before merging

This flips the xAI default from grok-4 to grok-4.5 (also in internal/providermodelcatalog/catalog.go). If grok-4.5 is not served by api.x.ai (or the ID differs, e.g. grok-4-5), every default xAI request 404s on first use. grok-4 is retained as "previous flagship" so a revert is trivial. Could not verify without API access.

[lfaoro]

Thanks @gnanam1990 for the review.

grok-4.5 is the live model ID on api.x.ai.
https://docs.x.ai/developers/models

[lfaoro]

xAI /models endpoint is gated by token/credits but you can see the model ID via openrouter open endpoint.

curl -s https://openrouter.ai/api/v1/models \
  | jq '.data[] | select(.id | startswith("x-ai")) | {id, name}'

[Vasanthdev2004]

Looks good bumping the xAI default to grok-4.5 across the provider descriptor, the curated model list (grok-4 retained as "previous flagship"), and the onboarding test fixture. grok-4.5 is the real publicly released xAI model (announced July 8, 2026, id grok-4.5), build/vet/gofmt are clean on the touched packages, and the providercatalog, providermodelcatalog, and xAI/onboarding tui tests pass. No attribution-header or trust-anchor concerns on this one. Approving.