feat(minimax): refresh Token Plan usage display

README.md

@@ -36,7 +36,7 @@ OpenUsage lives in your menu bar and shows you how much of your AI coding subscr
 - [**JetBrains AI Assistant**](docs/providers/jetbrains-ai-assistant.md) / quota, remaining
 - [**Kiro**](docs/providers/kiro.md) / credits, bonus credits, overages
 - [**Kimi Code**](docs/providers/kimi.md) / session, weekly
-- [**MiniMax**](docs/providers/minimax.md) / coding plan session
+- [**MiniMax**](docs/providers/minimax.md) / token plan model-calls, CN TTS/image buckets
 - [**OpenCode Go**](docs/providers/opencode-go.md) / 5h, weekly, monthly spend limits
 - [**Windsurf**](docs/providers/windsurf.md) / prompt credits, flex credits
 - [**Z.ai**](docs/providers/zai.md) / session, weekly, web searches

docs/providers/minimax.md

@@ -8,6 +8,8 @@
 - **Endpoint:** `GET https://api.minimax.io/v1/api/openplatform/coding_plan/remains`
 - **Auth:** `Authorization: Bearer <api_key>`
 - **Window model:** dynamic rolling 5-hour limit (per MiniMax Coding Plan docs)
+- **Display note:** OpenUsage renders the main `Session` line as a percentage (`0`-`100`) so it visually aligns with other providers (claude/codex), and renders companion resource buckets (TTS HD/Turbo, `image-01`, etc.) as raw counts (`chars`/`images`).
+- **CN note:** current CN docs use `https://www.minimaxi.com/v1/api/openplatform/coding_plan/remains`.
 
 ## Authentication
 
@@ -44,6 +46,7 @@ Fallbacks:
 
 When the selected region is `CN`, requests use:
 
+- `https://www.minimaxi.com/v1/api/openplatform/coding_plan/remains`
 - `https://api.minimaxi.com/v1/api/openplatform/coding_plan/remains`
 - `https://api.minimaxi.com/v1/coding_plan/remains`
 
@@ -61,25 +64,48 @@ Expected payload fields:
 
 ## Usage Mapping
 
-- Treat `current_interval_usage_count` as remaining prompts (MiniMax remains API behavior).
+- Treat `current_interval_usage_count` as the remaining raw session/resource count returned by the remains API.
+- For the main text `Session` line, OpenUsage emits a percentage (`0`-`100`) computed from the raw remains numbers (no prompt conversion). Companion resource buckets keep their raw counts (`chars` / `images`).
 - If only remaining aliases are provided, compute `used = total - remaining`.
 - If explicit used-count fields are provided, prefer them.
-- Plan name is taken from explicit plan/title fields when available.
-- If plan fields are missing in GLOBAL mode, infer plan tier from known limits (`100/300/1000/2000` prompts or `1500/4500/15000/30000` model-call equivalents).
-- If plan fields are missing in CN mode, infer only exact known CN limits (`600/1500/4500` model-call counts).
+- Plan name is taken from explicit plan/title fields when available, and normalized to a shared six-plan naming scheme:
+  - `Starter`
+  - `Plus`
+  - `Max`
+  - `Plus-High-Speed`
+  - `Max-High-Speed`
+  - `Ultra-High-Speed`
+- If plan fields are missing, infer the plan tier from the current Token Plan quota table for the selected region:
+  - `GLOBAL` raw `model-calls`: `1500 => Starter`, `4500 => Plus`, `15000 => Max`, `30000 => Ultra-High-Speed`
+  - `CN` raw `model-calls`: `600 => Starter`, `1500 => Plus`, `4500 => Max`, `30000 => Ultra-High-Speed`
+- For overlapping middle tiers, the plugin also inspects companion daily quotas when present to disambiguate `Standard` vs `High-Speed`:
+  - `GLOBAL 4500`: `image-01 50` or `Speech 2.8 4000` => `Plus`; `image-01 100` or `Speech 2.8 9000` => `Plus-High-Speed`
+  - `GLOBAL 15000`: `image-01 120` or `Speech 2.8 11000` => `Max`; `image-01 200` or `Speech 2.8 19000` => `Max-High-Speed`
+  - `CN 1500`: `image-01 50` or `speech-hd 4000` => `Plus`; `image-01 100` or `speech-hd 9000` => `Plus-High-Speed`
+  - `CN 4500`: `image-01 120` or `speech-hd 11000` => `Max`; `image-01 200` or `speech-hd 19000` => `Max-High-Speed`
+- If those companion quotas are absent or conflicting, the plugin falls back to the coarse family label (`Plus` / `Max`) instead of guessing.
+- Additional `model_remains[]` companion resource buckets are rendered as separate daily detail lines in both `GLOBAL` and `CN` mode, for example `speech-hd` (`Text to Speech HD`) or `image-01`.
 - Use `end_time` for reset timestamp when present.
 - Fallback to `remains_time` when `end_time` is absent.
 - Use `start_time` + `end_time` as `periodDurationMs` when both are valid.
+- Non-session companion resource lines use a daily window when only `remains_time` is present.
+- Prompt-based marketing copy is ignored by the plugin; all inference is based on raw remains quotas and companion resource buckets.
+- Official package tables used for this split, checked on 2026-05-18:
+  - Global: <https://platform.minimax.io/docs/token-plan/intro>
+  - CN: <https://platform.minimaxi.com/docs/token-plan/intro>
 
 ## Output
 
 - **Plan**: best-effort from API payload (normalized to concise label, with ` (CN)` or ` (GLOBAL)` suffix)
 - **Session** (overview progress line):
   - `label`: `Session`
-  - `format`: count (`prompts`)
-  - `used`: computed used prompts
-  - `limit`: total prompt limit for current window
+  - `format`: percent
+  - `used`: computed session usage percentage (`0`-`100`) derived from raw remains data
+  - `limit`: `100`
   - `resetsAt`: derived from `end_time` or `remains_time`
+- **Extra resources** (detail progress lines when present in either region):
+  - `Text to Speech HD` / `Text to Speech Turbo`: count (`chars`)
+  - `Image Generation` / `image-01`: count (`images`)
 
 ## Errors