feat: add grok-4.5 model alias and pricing

[Copilot]

grok-4.5 (xAI via Copilot) had zero alias coverage and was absent from pricing data despite being actively served with full billing metadata.

Changes

• pkg/workflow/data/model_aliases.json — new grok alias family:

  "grok": ["copilot/*grok*", "openai/*grok*"]

• pkg/cli/data/models.json + actions/setup/js/models.json (kept byte-identical):

  • grok-4.5 under github-copilot provider
  • github/grok-4.5 mirror under openai provider
  • Pricing sourced from Copilot SDK billing.tokenPrices: input $0.20/M, output $0.60/M, cache-read $0.05/M
  • provider_type: "openai", wire_api: "completions" — consistent with other non-native Copilot-routed models (kimi, raptor-mini, gemini variants)

• Fixes[model-inventory] Model alias inventory update - 2026-07-30 #49014

[github-actions]

Triage Summary

• Category: feature
• Risk: low
• Priority Score: 36/100 (impact 18 + urgency 8 + quality 10)
• Recommended Action: defer

Adds grok-4.5 model alias/pricing. Still draft, small and low-risk; not urgent.

Generated by 🔧 PR Triage Agent · auto · 54.9 AIC · ⌖ 5.64 AIC · ⊞ 7.9K · ◷

[Copilot]

Pull request overview

Adds Grok 4.5 alias resolution and pricing metadata.

Changes:

• Adds the grok alias family.
• Adds Grok 4.5 pricing to both mirrored catalogs.
• Configures OpenAI-compatible completions routing.

Show a summary per file

File	Description
pkg/workflow/data/model_aliases.json	Adds Grok alias patterns.
pkg/cli/data/models.json	Adds CLI pricing metadata.
actions/setup/js/models.json	Mirrors runtime pricing metadata.

Review details

Tip

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

Comments suppressed due to low confidence (2)

actions/setup/js/models.json:950

• The OpenAI-provider mirror also lacks Grok's documented 2× long-context tier above 200K tokens. As a result, runtime cost reporting for github/grok-4.5 will use base rates and undercount these requests by 50%.

 "input": "2e-07",
"output": "6e-07",
"cache_read": "5e-08"

pkg/cli/data/models.json:950

• The github/grok-4.5 mirror also omits the documented 2× pricing tier above 200K tokens, so long-context usage through this provider path will be billed at half the stated rate. Please add threshold-aware pricing consistently for this entry.

 "input": "2e-07",
"output": "6e-07",
"cache_read": "5e-08"

• Files reviewed: 3/3 changed files
• Comments generated: 3
• Review effort level: Medium

[pelikhan]

@copilot run pr-finisher skill

[pelikhan]

@copilot resolve the merge conflicts on this branch.

[Copilot]

@copilot run pr-finisher skill

Done. Added grok test coverage in TestBuiltinModelAliases (commit test: add grok alias test coverage):

• Added grok to expectedFamilies and vendorFamilies
• Added explicit pattern assertion: assert.Equal(t, []string{"copilot/*grok*", "openai/*grok*"}, aliases["grok"], ...)

Regarding the threshold pricing feedback (2× tier above 200K tokens): neither the Go nor JavaScript cost calculators currently support token-count-based pricing tiers — all models in the catalog use flat per-token rates. Adding threshold pricing would be a new feature spanning model_costs.go, model_costs.cjs, and the JSON schema, which is beyond the scope of this data-entry PR. I've tracked this as a follow-up to implement tiered pricing support across the catalog.

[Copilot]

@copilot resolve the merge conflicts on this branch.

Resolved. Merged origin/main, accepted main's versions for the 8 conflicting lock files (all generated; our branch doesn't touch any .md workflow sources), then ran make recompile to resync all 266 lock files with their markdown sources.

[github-actions]

🎉 This pull request is included in a new release.

Release: v0.84.1