Add Kimi K3 to the Copilot model inventory

[Copilot]

The models.dev catalog lists kimi-k3, but the local Copilot pricing inventory did not. Existing wildcard aliases already cover the model.

• Model inventory

  • Add kimi-k3 to both synchronized catalogs.
  • Record per-token pricing for $3 input, $15 output, and $0.30 cache read per million tokens.
  • Configure OpenAI-compatible completions metadata.

• Coverage

  • Add focused pricing lookup assertions.
  • Add a patch changeset for the inventory update.

• Fixes [model-inventory] Model alias inventory update - 2026-08-10 #51656

[github-actions]

🧠 Matt Pocock Skills Reviewer has completed the skills-based review. ✅

🧠 Reviewed using Matt Pocock's skills by Matt Pocock Skills Reviewer

[github-actions]

✅ Test Quality Sentinel completed test quality analysis.

🧪 Test quality analysis by Test Quality Sentinel

[github-actions]

✅ PR Code Quality Reviewer completed the code quality review.

Warning

Threat Detection Engine Failure — The analysis engine could not complete. This is a tooling failure, not a security finding.

What happened

The threat detection engine failed to produce results.

Review the workflow run logs for details.

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• api.individual.githubcopilot.com

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "api.individual.githubcopilot.com"

See Network Configuration for more information.

🔎 Code quality review by PR Code Quality Reviewer

[github-actions]

✅ Ponytail Reviewer completed successfully!

Lean already. Ship.

Generated by Ponytail Reviewer for #51938

[github-actions]

✅ Design Decision Gate 🏗️ completed the design decision gate check.

No ADR enforcement needed: PR does not have the 'implementation' label and has ≤100 new lines of code in business logic directories (17 additions detected).

🏗️ ADR gate enforced by Design Decision Gate 🏗️

[Copilot]

Pull request overview

Adds kimi-k3 pricing metadata to the synchronized model catalogs. The changes are consistent, but the PR remains WIP with final validation and security checks incomplete.

Changes:

• Adds Copilot pricing and API metadata for kimi-k3.
• Tests embedded pricing lookup.
• Documents the catalog update.

Show a summary per file

File	Description
pkg/cli/model_costs_test.go	Tests kimi-k3 pricing lookup.
pkg/cli/data/models.json	Adds the CLI catalog entry.
actions/setup/js/models.json	Mirrors the catalog entry for setup runtime.
.changeset/patch-model-inventory-2026-08-10.md	Records the patch release change.

Review details

Tip

Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

• Files reviewed: 4/4 changed files
• Comments generated: 0
• Review effort level: Balanced

[github-actions]

Excellent work on the model catalog update! 🎉 This PR adds the kimi-k3 model to both synchronized inventories (actions/setup/js/models.json and pkg/cli/data/models.json) with appropriate pricing tiers, directly addressing the gap identified in the daily model inventory check (#51656).

The PR is well-focused on a single responsibility, includes test coverage for the model cost logic, and is properly documented with a linked issue. The changes align perfectly with the project's model maintenance workflow.

✅ Ready for review and merge!

Generated by ✅ Contribution Check · auto · 49.8 AIC · ⌖ 3.79 AIC · ⊞ 8.8K · ◷

[github-actions]

The kimi-k3 model addition looks correct: pricing values match the changeset description, both models.json copies are updated in sync, and the test exercises all three cost fields. No issues found.

🧵 Reviewed using Impeccable skills by Impeccable Skills Reviewer · sonnet46 · 14.8 AIC · ⌖ 6.17 AIC · ⊞ 5.4K

[github-actions]

Skills-Based Review 🧠

Applied /grill-with-docs — one inaccuracy in the changeset description to fix.

📋 Key Themes & Highlights

Key Themes

• Changeset inaccuracy: The changeset claims the existing kimi alias covers the new model, but no such alias exists in models.json.

Positive Highlights

• ✅ Test added alongside the data change, asserting all three cost dimensions
• ✅ Both synchronized inventories (actions/setup/js/models.json and pkg/cli/data/models.json) updated identically
• ✅ Changeset included with correct pricing figures

🧠 Reviewed using Matt Pocock's skills by Matt Pocock Skills Reviewer · sonnet46 · 26.9 AIC · ⌖ 6.96 AIC · ⊞ 7.1K
_{Comment /matt to run again}

[github-actions]

[/grill-with-docs] The changeset states "the existing kimi alias already covers the new model" but no kimi alias exists in models.json — only kimi-k2.7-code alongside the new kimi-k3. This claim is inaccurate and could mislead reviewers.

💡 Suggestion

Remove the alias claim. Corrected text:

Model alias inventory update for 2026-08-10: add `kimi-k3` ($3/1M input, $15/1M output, $0.30/1M cache read) to the `github-copilot` provider in `models.json`.

@copilot please address this.

[github-actions]

🧪 Test Quality Sentinel Analysis

PR: #51938 — "[WIP] Add kimi-k3 model to models.json"
Test Quality Score: 70/100 ⚠️ Acceptable

Summary

This PR adds the kimi-k3 model to the model inventory and includes one new test function TestFindKimiK3Pricing that validates the model's pricing is correctly registered. The test is well-structured with proper assertions but covers only the happy path.

Test Coverage

Metric	Value
New test functions	1
Test file	pkg/cli/model_costs_test.go
Assertions per test	4 (1× require.True, 3× assert.InDelta)
Test:Production lines ratio	~1:1 ✅ (8 test lines vs ~18 production changes)
Design tests	100%
Edge case coverage	0% (happy path only)

New Test Details

TestFindKimiK3Pricing — Model pricing lookup

Classification:

• Design invariant: behavioral_contract — verifies that new model pricing is accessible
• Value if deleted: high_value — regression would allow model addition without pricing verification
• Contract level: design_test — tests public findModelPricing() lookup contract

Implementation:

func TestFindKimiK3Pricing(t *testing.T) {
	pricing, ok := findModelPricing("github-copilot", "kimi-k3")
	require.True(t, ok)
	assert.InDelta(t, 0.000003, pricing["input"], 1e-12)
	assert.InDelta(t, 0.000015, pricing["output"], 1e-12)
	assert.InDelta(t, 0.0000003, pricing["cache_read"], 1e-12)
}

Strengths:

• ✅ Proper build tag ((go/redacted):build !integration)
• ✅ High-precision price assertions (1e-12 tolerance)
• ✅ Verifies all pricing dimensions (input, output, cache_read)
• ✅ Distinct from existing TestFindModelPricing test

Notes:

• Covers happy path only; negative case (missing model) already tested by TestFindModelPricing
• Data-driven test: model pricing is config-based, so single-path verification is appropriate for inventory additions

Quality Scoring Breakdown

Design tests (1/1 = 100%):       40 pts ✅
Edge cases (0/1 = 0%):           0 pts
Duplicate penalty:               0 pts ✅
Inflation ratio (<2:1):          10 pts ✅
─────────────────────────────────
Total:                           70/100

Status

✅ No violations — Build tags present, no mock library usage, appropriate assertion pattern
✅ Healthy test:code ratio — 1:1 inflation (well below 2:1 threshold)
⚠️ Happy path only — Single-path test is acceptable for data-driven model inventory additions

Recommendation: APPROVE — This is a valid behavioral contract test appropriate for model inventory updates.

🧪 Test quality analysis by Test Quality Sentinel · haiku45 · 18.4 AIC · ⌖ 4.24 AIC · ⊞ 7.6K · ◷
_{Comment /review to run again}

[github-actions]

✅ Test Quality Sentinel: 70/100. 100% design tests (threshold: ≤30% implementation tests). No violations detected.

[github-actions]

🎉 This pull request is included in a new release.

Release: v0.86.2