Feature Request: Support Xiaomi MiMo Token Plan API Endpoint & Pricing Model

[springeye]

Is your feature request related to a problem? Please describe.

CodeWhale already ships a xiaomi/mimo provider for the standard MiMo API (pay-as-you-go), which works well for on-demand usage. However, Xiaomi's newly launched Token Plan subscription model (Lite / Standard / Pro / Max tiers at ¥39–¥659/month) uses a separate API endpoint and authentication scheme that is currently not supported.

Users who purchase a Token Plan subscription cannot use their Token Plan credentials with the existing MiMo provider — the base URL, auth flow, and model resolution paths differ from the standard API. This forces Token Plan subscribers to either:

• Manually configure a custom OpenAI-compatible endpoint (losing provider-specific optimizations like model lists, rate-limit awareness, and cost tracking), or
• Continue paying for pay-as-you-go even though they've already subscribed to a Token Plan.

Describe the solution you'd like

Extend the existing xiaomi/mimo provider to support the Token Plan API as a sub-mode or alternative endpoint:

1. Token Plan API endpoint: Add the official Token Plan base URL as an alternative endpoint option alongside the standard API endpoint
2. Auth credentials: Support Token Plan-specific API keys (issued from the Token Plan console at platform.xiaomimimo.com/token-plan)
3. Mode switch: Allow users to choose between "Standard (pay-as-you-go)" and "Token Plan (subscription)" via config — e.g. a config field like provider.xiaomi.mode = "token-plan"
4. Model resolution: Token Plan subscribers get access to MiMo-V2.5-Pro, MiMo-V2.5, and TTS models — auto-discover or list these models specifically for Token Plan mode
5. Cost tracking: Since Token Plan uses a credit/point system rather than per-token billing, the cost display should reflect remaining credits rather than token-based cost
6. Rate-limit awareness: Respect Token Plan tier-specific rate limits

Describe alternatives you've considered

• Manual OpenAI-compatible endpoint override: Already possible via custom config, but loses CodeWhale's model list discovery, rate-limit tracking, and cost accounting features
• MCP proxy bridge: A custom MCP server could bridge the Token Plan API, but adds unnecessary complexity and maintenance burden
• OpenRouter fallback: OpenRouter carries MiMo models but routes through standard API pricing, negating the Token Plan subscription benefit

None of these provide the same UX as a first-class Token Plan mode within the existing MiMo provider.

Additional context

• CodeWhale already has excellent MiMo standard API support — this request is purely about adding the Token Plan variant as a sub-mode
• Xiaomi MiMo Token Plan official site: https://platform.xiaomimimo.com/token-plan
• MiMo API docs: https://platform.xiaomimimo.com/docs
• Token Plan has seen rapid adoption since launch, and many Chinese developers now access MiMo models exclusively through Token Plan subscriptions
• I'm happy to help test Token Plan integration once a prototype is available

[github-actions]

Thanks @springeye for the report.

This repository is currently observing a maintainer-managed contribution gate in dry-run mode, so this issue is staying open. When enforcement is enabled, issues from contributors who are not listed in .github/APPROVED_CONTRIBUTORS will be closed automatically.

Please read CONTRIBUTING.md for the expected issue shape. A maintainer can grant issue access by commenting /lgtmi on an issue.

[xyuai]

Opened PR #2627 for this.

What it adds:

• [providers.xiaomi_mimo].mode = "token-plan" plus token-plan-sgp / token-plan-ams cluster aliases.
• Token Plan tp-* key env aliases: XIAOMI_MIMO_TOKEN_PLAN_API_KEY and MIMO_TOKEN_PLAN_API_KEY.
• api-key authentication for the official Token Plan hosts while preserving Authorization: Bearer for the standard MiMo endpoint.
• MiMo chat / ASR / TTS model alias normalization and updated provider docs / examples.

Validated with:

• cargo test -p codewhale-config xiaomi_mimo
• cargo test -p codewhale-secrets xiaomi_mimo_env_aliases_resolve
• cargo test -p codewhale-tui --bin codewhale-tui xiaomi_mimo

Note: I wired the documented Token Plan endpoint/auth/model path. Xiaomi currently documents credits and rate-limit quotas through the Token Plan docs/console, but I did not find a documented balance API endpoint to poll for live remaining-credit telemetry yet.

[Hmbown]

Thanks @springeye. This is a useful tracking request, especially because Token Plan appears to change more than just the base URL: auth, model availability, quota/cost display, and rate limits all need to be represented correctly.

I am leaving this outside the v0.8.52 corrective release because it needs verified endpoint/auth behavior and probably a test account before we make it first-class. For now the safest path is to track it as a MiMo provider follow-up rather than squeeze an unverified mode into a stabilization release.

[Hmbown]

Status update: the Token Plan endpoint/auth/env-mode slice from #2627 landed via #2800 and is included in green #2762 at 125612575a3aaeb81c3a21456161716ad9bfc8ec. CodeWhale now supports mode / XIAOMI_MIMO_MODE / MIMO_MODE for token-plan-sgp, token-plan-cn, token-plan-ams, and pay-as-you-go, and keeps Token Plan tp-* keys on api-key auth while pay-as-you-go uses standard MiMo keys.

Thanks @springeye for the request and @xyuai for the implementation work. Leaving this issue open for the remaining model-list, cost/credit display, and rate-limit acceptance criteria.