feat(xai): add grok-4.3 model

[steebchen]

Summary

• Add grok-4.3 to the xAI provider in packages/models
• Includes 1M context window, vision, reasoning, tools, JSON output, and streaming support
• Tiered pricing: $1.25/$2.50 per 1M input/output tokens up to 200K, doubled above

Test plan

• TEST_MODELS="xai/grok-4-3" pnpm test:e2e — all 74 tests pass
• pnpm format

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features
  • Added support for Grok 4.3 model with advanced capabilities including streaming, vision analysis, reasoning, tool usage, and structured output. Features an expanded context window for handling larger documents and conversations.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 5ac38f21-8b0d-4c83-bc8b-d4a946538ac1

📥 Commits

Reviewing files that changed from the base of the PR and between df80555 and 0607ced.

📒 Files selected for processing (1)

• packages/models/src/models/xai.ts

---

Walkthrough

Adds a new XAI model definition (grok-4-3) to the xaiModels array with metadata, token-based pricing tiers, and capability flags including streaming, vision, reasoning, tools, and JSON output support with a context window of 1,000,000 tokens.

Changes

Cohort / File(s)	Summary
XAI Model Addition packages/models/src/models/xai.ts	Added new model definition grok-4-3 with metadata (name, description, family, release date), tiered input/output pricing, cached input pricing, and capability flags (streaming, vision, reasoning, tools, JSON output).

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

Possibly related PRs

• feat(models): align xAI Grok tiered pricing with official rates #2037: Modifies the same XAI model configuration file to adjust model provider pricing and tiered pricing structures.

🚥 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 accurately summarizes the main change: adding a new xAI model (grok-4.3) to the models package, which is the core objective of the pull request.
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.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch add-grok-4.3

---

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

_{Review rate limit: 7/8 reviews remaining, refill in 7 minutes and 30 seconds.}

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

[Copilot]

Pull request overview

Adds the new grok-4-3 model definition to the xAI model catalog in packages/models, including capabilities (1M context, vision, reasoning, tools, JSON output, streaming) and tiered pricing metadata.

Changes:

• Add xai/grok-4-3 model entry with pricing tiers (<=200K vs >200K tokens).
• Mark the model as streaming/vision/reasoning/tools/JSON-output capable.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

supportedParameters omits reasoning_effort, but prepareRequestBody only forwards reasoning_effort when the provider mapping’s supportedParameters is empty or explicitly includes it. As written, reasoning_effort will be silently dropped for this model (and any other model using xaiSupportedParamsNoFreqPresence). Add reasoning_effort to the supported parameter list for grok-4-3 (or to xaiSupportedParamsNoFreqPresence) so clients can control reasoning effort as intended.