fix(cursor): expose Grok 4.6 xhigh Fast by jbaehova · Pull Request #1615 · lidge-jun/opencodex

jbaehova · 2026-08-13T13:41:31Z

Summary

expose Cursor Grok 4.6's observed low / medium / high / xhigh ladder for both regular and Fast picker rows
preserve Cursor's existing parameterized Fast wire contract: send base model grok-4.6 with separate effort=xhigh and fast=true requested_model parameters
cover live discovery identities, picker effort metadata, regular request ids, Fast request parameters, and localized Cursor documentation

This is the Cursor-only follow-up requested when #1591 was closed as superseded. The direct xAI catalog half has already landed in #1593; this PR does not modify src/providers/registry.ts or any xAI routing behavior.

Root cause

The Cursor Grok 4.6 fallback rows were added by mirroring Grok 4.5's three-tier ladder before Cursor exposed its final lineup. Cursor now advertises cursor-grok-4.6-xhigh and cursor-grok-4.6-xhigh-fast, but OpenCodex still capped both rows at high. That hid Extra High from the picker and clamped explicit xhigh requests.

Fast requests remain parameterized rather than sending the flattened discovery slug: grok-4.6 plus effort=xhigh and fast=true. The flattened cursor-grok-4.6-xhigh-fast id is availability evidence only.

Verification

Cursor CLI account lineup includes cursor-grok-4.6-xhigh and cursor-grok-4.6-xhigh-fast
live Cursor request grok-4.6[effort=xhigh,fast=true] completed successfully
focused Cursor regression tests on the rebased head — 22 passed, 0 failed
full repository suite on the rebased head with Bun 1.3.14 — 11,570 passed, 8 skipped, 0 failed
storage-policy tests — 9 passed, 0 failed
api-usage tests — 16 passed, 0 failed
GUI tests — 790 passed, 0 failed
TypeScript typecheck and GUI lint — passed
privacy scan — passed
docs build — 265 pages built successfully
git diff --check — passed

Checklist

Scope stays focused and avoids unrelated cleanup.
Docs or release notes were updated when needed.
Security-sensitive changes were reviewed; this changes Cursor capability metadata and model selection only, with no credential or authentication behavior changes.

Review readiness checklist

This PR stays in draft until every box below is ticked. Tick all four boxes once the requirements are met:

All CI tests are green on my local testing.
I pushed my PR to the latest dev commit.
I resolved all correct Codex and CodeRabbit findings.
My PR is ready for review.

Summary by CodeRabbit

New Features
- Added Grok 4.6 and Grok 4.6 Fast alongside Grok 4.5 variants.
- Added the xhigh reasoning tier for Grok 4.6.
- Improved model discovery for regular and Fast variants, including context details.
Documentation
- Updated provider and model guides across supported languages.
- Documented Fast request behavior, reasoning tiers, and model availability identifiers.
Tests
- Expanded coverage for Grok 4.6 discovery, reasoning levels, context limits, and Fast requests.

github-actions · 2026-08-13T13:41:45Z

✅ Deterministic PR hygiene checks passed.

coderabbitai · 2026-08-13T13:41:57Z

Important

Review skipped

No new commits to review since the last review.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 39a9d618-4b19-4674-a4b9-5875611de9e7

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

🔍 Trigger review

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: bd0b3b2d-c9b9-4d53-b4f9-b4427c738142

📥 Commits

Reviewing files that changed from the base of the PR and between efe7f44 and b9568aa.

📒 Files selected for processing (1)

tests/cursor-effort-suffix.test.ts

📝 Walkthrough

Walkthrough

Cursor support now includes Grok 4.6 regular and Fast models, the xhigh effort tier, version-specific Fast request parameters, expanded discovery handling, tests, and localized documentation.

Changes

Cursor Grok model support

Layer / File(s)	Summary
Routing and effort mappings `src/adapters/cursor/effort-map.ts`, `src/adapters/cursor/request-builder.ts`, `structure/04_transports-and-sidecars.md`	Grok 4.6 adds `xhigh` support. Fast requests use the matching base model with `effort`, `fast=true`, and `requested_model`. Discovery handles flattened variants for both Grok versions.
Discovery and effort validation `tests/cursor-discovery.test.ts`, `tests/cursor-effort-suffix.test.ts`	Tests cover Grok 4.6 regular and Fast IDs, `xhigh` filtering, context-window inference, effort ladders, suffix mapping, and Fast parameter translation.
Catalog and provider documentation `docs-site/src/content/docs/guides/`, `docs-site/src/content/docs/ja/guides/`, `docs-site/src/content/docs/ko/guides/`, `docs-site/src/content/docs/ru/guides/`, `docs-site/src/content/docs/zh-cn/guides/`, `docs-site/src/content/docs/zh-tw/guides/`	Localized documentation lists Grok 4.5/4.6 regular and Fast variants, effort tiers, Fast wire parameters, and discovery-only flattened IDs.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Mergeability Score: ⚪ Minimal · up to b9568

This localized change exposes Cursor Grok 4.6 xhigh options while preserving the existing Fast request contract, with the reported focused, full-suite, typecheck, lint, documentation, and privacy checks passing; no actionable merge-blocking risk remains.

Sequence Diagram(s)

sequenceDiagram
  participant Client
  participant CursorDiscovery
  participant CursorEffortMap
  participant CursorRequestBuilder
  Client->>CursorDiscovery: discover Grok 4.6 variants
  CursorDiscovery->>CursorEffortMap: map effort suffixes
  CursorEffortMap-->>CursorDiscovery: return supported effort tiers
  Client->>CursorRequestBuilder: select Grok 4.6 Fast
  CursorRequestBuilder->>CursorRequestBuilder: set base model, effort, fast=true, requested_model

Possibly related PRs

lidge-jun/opencodex#1591: Updates the same Cursor Grok 4.6 effort mappings, request parameters, discovery behavior, documentation, and tests.
lidge-jun/opencodex#1547: Shares Cursor Grok 4.6 routing, effort mapping, discovery, and test coverage.
lidge-jun/opencodex#1228: Overlaps in Cursor effort mapping, Fast request handling, and related discovery tests.

Suggested labels: enhancement

Suggested reviewers: ingwannu, lidge-jun

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
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.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the main change: exposing Grok 4.6 xhigh support for Cursor Fast models.

✨ 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

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

github-actions · 2026-08-13T13:41:58Z

✅ READY

all PR quality gates passed; the review readiness checklist is complete.

Review readiness checklist

✅ All CI tests are green on my local testing.
✅ I pushed my PR to the latest dev commit.
✅ I resolved all correct Codex and CodeRabbit findings.
✅ My PR is ready for review.

✅ 4/4 boxes ticked.

This pull request is already Ready for Review.
The review-ready label marks this PR as ready; review automation runs independently. If no CodeRabbit review appears, comment @coderabbitai review to request one.
Maintainers: @lidge-jun @Ingwannu @Wibias

coderabbitai · 2026-08-13T13:45:28Z

Tip

For best results, initiate chat on the files or code changes.

@github-actions[bot] PR #1615 is marked ready for review at efe7f440923f38100fda02db58d6605ef418e204.

I am replying to the latest status only. Earlier comment context is unavailable.

🧠 Learnings used

Learnt from: Wibias
Repo: lidge-jun/opencodex PR: 986
File: .github/workflows/enforce-pr-target.yml:631-636
Timestamp: 2026-08-04T06:41:30.089Z
Learning: For the PR readiness gate in `.github/workflows/enforce-pr-target.yml`, the required scope is to bind contributor checklist completion and maintainer notification to the exact PR head SHA, then reset them after a new head commit. Maintainer approval staleness is separate from this gate and should be enforced by GitHub branch protection, such as `dismiss-stale-reviews` or `require_last_push_approval`, when configured for `dev`.

_{You are interacting with an AI system.}

coderabbitai · 2026-08-13T13:45:40Z

@github-actions[bot] The readiness checklist is complete. I will review pull request #1615 at efe7f440923f38100fda02db58d6605ef418e204.

🧠 Learnings used

Learnt from: Wibias
Repo: lidge-jun/opencodex PR: 986
File: .github/workflows/enforce-pr-target.yml:631-636
Timestamp: 2026-08-04T06:41:30.089Z
Learning: For the PR readiness gate in `.github/workflows/enforce-pr-target.yml`, the required scope is to bind contributor checklist completion and maintainer notification to the exact PR head SHA, then reset them after a new head commit. Maintainer approval staleness is separate from this gate and should be enforced by GitHub branch protection, such as `dismiss-stale-reviews` or `require_last_push_approval`, when configured for `dev`.

⚠️ Action not completed

Already reviewed.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

Wibias · 2026-08-13T17:07:46Z

Thanks for following up with this Cursor-only version. This addresses the main concern I had on #1578: the Fast xhigh behavior now has both live evidence and a regression that pins the actual wire contract (grok-4.6 + effort=xhigh + fast=true) instead of relying on the flattened discovery slug.

The scope also looks right now: Grok 4.5 stays capped at high, only Cursor Grok 4.6 gains xhigh, and the already-landed xAI work remains separate.

I don’t see a code blocker here. Please rebase onto latest dev and let the repository CI actually run on the rebased head. One small non-blocking improvement would be to add an explicit no-effort assertion for both regular and Fast Grok 4.6, so the fallback/default behavior stays intentional if the ladder changes again.

After rebase + green CI, this looks good to merge.

coderabbitai · 2026-08-13T22:17:05Z

Tip

For best results, initiate chat on the files or code changes.

@github-actions[bot] The readiness checklist is complete for efe7f440923f38100fda02db58d6605ef418e204.

This head SHA was already reviewed. No additional incremental review is available until the PR receives a new commit.

🧠 Learnings used

Learnt from: Wibias
Repo: lidge-jun/opencodex PR: 986
File: .github/workflows/enforce-pr-target.yml:631-636
Timestamp: 2026-08-04T06:41:30.089Z
Learning: For the PR readiness gate in `.github/workflows/enforce-pr-target.yml`, the required scope is to bind contributor checklist completion and maintainer notification to the exact PR head SHA, then reset them after a new head commit. Maintainer approval staleness is separate from this gate and should be enforced by GitHub branch protection, such as `dismiss-stale-reviews` or `require_last_push_approval`, when configured for `dev`.

_{You are interacting with an AI system.}

lidge-jun · 2026-08-13T23:46:23Z

I reviewed the full diff, the Cursor request-builder and protobuf encoder paths, the discovery callers, the tests, the documentation updates, and the existing review discussion.

The implementation is appropriately scoped. src/adapters/cursor/effort-map.ts:38-40 adds xhigh only for Grok 4.6, while src/adapters/cursor/request-builder.ts:188-196 retains the established Fast wire contract: base grok-4.6 plus separate effort and fast=true requested-model parameters. The added discovery, ladder, regular-wire-id, and Fast-parameter tests cover the intended model-selection behavior, and the user-facing documentation is synchronized across the supported locales.

I agree with the prior maintainer review: this needs a rebase and real CI before merge. The PR is still draft and its head is 35 commits behind current dev. Its status rollup contains policy/hygiene/CodeRabbit results but no test-matrix result for a rebased SHA. Rebase onto latest dev, keep this exact narrow diff, mark it ready, and wait for the CI matrix to pass on that new head.

A non-blocking follow-up: an explicit no-reasoning assertion for both Grok 4.6 rows would keep their default-to-top-tier behavior intentional if the ladder changes later.

DISPOSITION: NEEDS-CHANGE

coderabbitai · 2026-08-14T00:01:07Z

Tip

For best results, initiate chat on the files or code changes.

@github-actions[bot] The readiness checklist is complete for efe7f440923f38100fda02db58d6605ef418e204.

This head SHA was already reviewed. No additional incremental review is available until the pull request receives a new commit.

🧠 Learnings used

Learnt from: Wibias
Repo: lidge-jun/opencodex PR: 986
File: .github/workflows/enforce-pr-target.yml:631-636
Timestamp: 2026-08-04T06:41:30.089Z
Learning: For the PR readiness gate in `.github/workflows/enforce-pr-target.yml`, the required scope is to bind contributor checklist completion and maintainer notification to the exact PR head SHA, then reset them after a new head commit. Maintainer approval staleness is separate from this gate and should be enforced by GitHub branch protection, such as `dismiss-stale-reviews` or `require_last_push_approval`, when configured for `dev`.

_{You are interacting with an AI system.}

coderabbitai · 2026-08-14T00:02:35Z

@github-actions[bot] The readiness checklist is complete for b9568aa8133098738f029bb447c5c295b8010ec3. I will review pull request #1615.

🧠 Learnings used

Learnt from: Wibias
Repo: lidge-jun/opencodex PR: 986
File: .github/workflows/enforce-pr-target.yml:631-636
Timestamp: 2026-08-04T06:41:30.089Z
Learning: For the PR readiness gate in `.github/workflows/enforce-pr-target.yml`, the required scope is to bind contributor checklist completion and maintainer notification to the exact PR head SHA, then reset them after a new head commit. Maintainer approval staleness is separate from this gate and should be enforced by GitHub branch protection, such as `dismiss-stale-reviews` or `require_last_push_approval`, when configured for `dev`.

⚠️ Action not completed

Already reviewed.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

coderabbitai · 2026-08-14T00:03:59Z

Tip

For best results, initiate chat on the files or code changes.

@github-actions The readiness gate is complete for b9568aa8133098738f029bb447c5c295b8010ec3. The review was already triggered for this head SHA.

🧠 Learnings used

Learnt from: Wibias
Repo: lidge-jun/opencodex PR: 986
File: .github/workflows/enforce-pr-target.yml:631-636
Timestamp: 2026-08-04T06:41:30.089Z
Learning: For the PR readiness gate in `.github/workflows/enforce-pr-target.yml`, the required scope is to bind contributor checklist completion and maintainer notification to the exact PR head SHA, then reset them after a new head commit. Maintainer approval staleness is separate from this gate and should be enforced by GitHub branch protection, such as `dismiss-stale-reviews` or `require_last_push_approval`, when configured for `dev`.

_{You are interacting with an AI system.}

jbaehova · 2026-08-14T00:05:39Z

Rebased onto the latest dev (8b1c6208) and force-updated the PR head to b9568aa8.

I also added the suggested explicit no-effort assertions for both regular and Fast Grok 4.6, pinning the defaults to cursor-grok-4.6-xhigh and grok-4.6 + effort=xhigh + fast=true respectively.

Validation on the rebased head:

focused Cursor regressions: 22 passed, 0 failed
full repository suite: 11,570 passed, 8 skipped, 0 failed
TypeScript typecheck, privacy scan, and GUI lint: passed
CodeRabbit: passed with no actionable comments or review threads

The PR is ready for review again. Cross-platform CI and React Doctor are currently awaiting maintainer approval to run on the forked head.

coderabbitai · 2026-08-14T00:06:20Z

@github-actions The readiness gate is complete for b9568aa8133098738f029bb447c5c295b8010ec3. I will review the current PR head.

🧠 Learnings used

Learnt from: Wibias
Repo: lidge-jun/opencodex PR: 986
File: .github/workflows/enforce-pr-target.yml:631-636
Timestamp: 2026-08-04T06:41:30.089Z
Learning: For the PR readiness gate in `.github/workflows/enforce-pr-target.yml`, the required scope is to bind contributor checklist completion and maintainer notification to the exact PR head SHA, then reset them after a new head commit. Maintainer approval staleness is separate from this gate and should be enforced by GitHub branch protection, such as `dismiss-stale-reviews` or `require_last_push_approval`, when configured for `dev`.

⚠️ Action not completed

Already reviewed.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

Wibias · 2026-08-14T02:21:49Z

Thanks for the follow-up and for tightening this up after review. This is useful because Cursor's Grok 4.6 support now matches the actual live lineup instead of silently stopping at high, while Fast keeps the correct parameterized wire contract rather than relying on the flattened discovery slug. The explicit default/no-effort coverage also makes that behavior much harder to regress if Cursor changes the ladder again.

CI is green on the rebased head, so I merged this. Thanks for the contribution.

jbaehova requested review from Ingwannu, Wibias and lidge-jun as code owners August 13, 2026 13:41

github-actions Bot added the bug Something isn't working label Aug 13, 2026

github-actions Bot added the review-ready label Aug 13, 2026

Wibias marked this pull request as draft August 13, 2026 22:16

Wibias removed the review-ready label Aug 13, 2026

fix(cursor): expose Grok 4.6 xhigh Fast

b9568aa

jbaehova force-pushed the agent/fix-cursor-grok-4-6-xhigh-fast branch from efe7f44 to b9568aa Compare August 14, 2026 00:00

github-actions Bot added the review-ready label Aug 14, 2026

github-actions Bot marked this pull request as ready for review August 14, 2026 00:02

Wibias merged commit c2eb678 into lidge-jun:dev Aug 14, 2026
30 checks passed

Conversation

jbaehova commented Aug 13, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Root cause

Verification

Checklist

Review readiness checklist

Summary by CodeRabbit

Uh oh!

github-actions Bot commented Aug 13, 2026

Uh oh!

coderabbitai Bot commented Aug 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review skipped

Walkthrough

Changes

Sequence Diagram(s)

❌ Failed checks (1 warning)

Uh oh!

github-actions Bot commented Aug 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ READY

Review readiness checklist

Uh oh!

coderabbitai Bot commented Aug 13, 2026

Uh oh!

coderabbitai Bot commented Aug 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Wibias commented Aug 13, 2026

Uh oh!

coderabbitai Bot commented Aug 13, 2026

Uh oh!

lidge-jun commented Aug 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai Bot commented Aug 14, 2026

Uh oh!

coderabbitai Bot commented Aug 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai Bot commented Aug 14, 2026

Uh oh!

jbaehova commented Aug 14, 2026

Uh oh!

coderabbitai Bot commented Aug 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Wibias commented Aug 14, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

jbaehova commented Aug 13, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Aug 13, 2026 •

edited

Loading

github-actions Bot commented Aug 13, 2026 •

edited

Loading

coderabbitai Bot commented Aug 13, 2026 •

edited

Loading

lidge-jun commented Aug 13, 2026 •

edited

Loading

coderabbitai Bot commented Aug 14, 2026 •

edited

Loading

coderabbitai Bot commented Aug 14, 2026 •

edited

Loading