feat(opencode): add modelBreakdowns to JSON output

[mjyoo2]

Summary

Add modelBreakdowns field to @ccusage/opencode JSON output for feature parity with ccusage (Claude Code).
Currently @ccusage/opencode only provides modelsUsed (model name array), while ccusage provides detailed modelBreakdowns with per-model token counts and costs.

Changes

• Add modelBreakdowns array to JSON output in daily, weekly, monthly, and session reports
• Each breakdown includes: modelName, inputTokens, outputTokens, cacheCreationTokens, cacheReadTokens, cost
• Backward compatible - modelsUsed array is preserved

Before/After

Before

{
  "daily": [{
    "date": "2026-01-20",
    "inputTokens": 228,
    "outputTokens": 346277,
    "totalCost": 95.38,
    "modelsUsed": ["claude-opus-4-5", "gpt-5.2-codex"]
  }]
}

After

{
  "daily": [{
    "date": "2026-01-20",
    "inputTokens": 228,
    "outputTokens": 346277,
    "totalCost": 95.38,
    "modelsUsed": ["claude-opus-4-5", "gpt-5.2-codex"],
    "modelBreakdowns": [
      {
        "modelName": "claude-opus-4-5",
        "inputTokens": 200,
        "outputTokens": 340000,
        "cacheCreationTokens": 7268400,
        "cacheReadTokens": 82593000,
        "cost": 94.50
      },
      {
        "modelName": "gpt-5.2-codex",
        "inputTokens": 28,
        "outputTokens": 6277,
        "cacheCreationTokens": 0,
        "cacheReadTokens": 0,
        "cost": 0.88
      }
    ]
  }]
}

Summary by CodeRabbit

• New Features
  • All reports enhanced with per-model breakdowns. Daily, weekly, monthly, and session reports now display comprehensive metrics for each AI model used, including input tokens, output tokens, cache tokens, and associated costs. Breakdowns are sorted by cost for easier identification of resource allocation.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR introduces per-model token and cost aggregation across four report command files (daily, monthly, session, weekly). Each command now accumulates metrics separately per model, creates sorted breakdowns, and extends output structures with a modelBreakdowns array. Synthetic model entries are skipped during aggregation.

Changes

Cohort / File(s)	Summary
Per-model aggregation across report commands apps/opencode/src/commands/daily.ts, apps/opencode/src/commands/monthly.ts, apps/opencode/src/commands/session.ts, apps/opencode/src/commands/weekly.ts	Introduces TokenStats and ModelBreakdown types to track per-model input/output/cache tokens and costs. Adds createModelBreakdowns helper to convert per-model aggregates into a sorted array by cost. Implements per-model aggregation logic using Map<string, TokenStats> while skipping synthetic entries. Extends report data structures with modelBreakdowns arrays and updates export flows accordingly.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

Possibly related PRs

• Add model breakdowns and usage details in usage reports #52: Adds identical per-model token and cost aggregation with modelBreakdowns types and logic across the same report commands.

Suggested reviewers

• ryoppippi

Poem

🐰 A rabbit's ode to model breakdowns:
Per-model tokens now shine so bright,
Aggregates sorted by cost in sight,
Synthetic entries hop away,
ModelBreakdowns guide the day! ✨

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes the primary change: adding modelBreakdowns to JSON output across opencode commands (daily, weekly, monthly, session).

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

✨ Finishing touches

• 📝 Generate docstrings

---

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 and usage tips.}