opencode-peak-scheduler 0.1.1 → 0.1.3

package/README.md

@@ -12,7 +12,7 @@ plugin API and the session-model lever were confirmed against that source (see
 
 | Window   | Model                         | When                                       |
 | -------- | ----------------------------- | ------------------------------------------ |
-| Peak     | `deepseek/deepseek-v4-flash`  | Mon–Fri 09:00–12:00 and 14:00–18:00        |
+| Peak     | `deepseek/deepseek-flash`  | Mon–Fri 09:00–12:00 and 14:00–18:00        |
 | Off-peak | `deepseek/deepseek-v4-pro`    | Everything else + weekends (Sat & Sun)     |
 
 Half-open intervals `[start, end)`. Weekday means Monday–Friday in the
@@ -41,7 +41,7 @@ adds one pseudo-model, `auto`, which becomes the default model:
         "baseURL": "https://api.deepseek.com",
         "apiKey": "…",
         "scheduler": {
-          "peakModel": "deepseek-v4-flash",
+          "peakModel": "deepseek-flash",
           "offModel": "deepseek-v4-pro",
           "timezone": "Asia/Shanghai",
           "autoID": "auto"
@@ -49,7 +49,7 @@ adds one pseudo-model, `auto`, which becomes the default model:
       },
       "models": {
         "deepseek-v4-pro": { "name": "DeepSeek-V4-Pro" },
-        "deepseek-v4-flash": { "name": "deepseek-v4-flash" },
+        "deepseek-flash": { "name": "deepseek-flash" },
         "auto": {
           "name": "Auto (peak-aware)",
           "limit": { "context": 1048576, "output": 393216 }
@@ -102,12 +102,12 @@ rate). Configure the **off-peak** per-1M-token prices on the model entries in
 `https://api-docs.deepseek.com/quick_start/pricing/`):
 
     "deepseek-v4-pro":  { "cost": { "input": 0.66, "output": 1.98, "cache": { "read": 0.022, "write": 0 } } },
-    "deepseek-v4-flash": { "cost": { "input": 0.22, "output": 0.66, "cache": { "read": 0.007, "write": 0 } } }
+    "deepseek-flash": { "cost": { "input": 0.15, "output": 0.6, "cache": { "read": 0.003, "write": 0 } } }
 
 `input` is the cache-miss price, `cache.read` the cache-hit price, and
 `cache.write` is 0 (DeepSeek doesn't charge for cache writes). DeepSeek's peak
 hours (`01:00–04:00` and `06:00–10:00` UTC) are exactly the plugin's peak
-windows, so during peak the plugin routes to `deepseek-v4-flash` at 2× and
+windows, so during peak the plugin routes to `deepseek-flash` at 2× and
 off-peak to `deepseek-v4-pro` at 1×. The estimate is accurate to the 30s check
 interval; in manual mode opencode prices the real model itself, so the plugin
 hides the line.
@@ -138,7 +138,7 @@ The sidebar also shows a `Balance:` line — the account's current DeepSeek
 balance, fetched via `GET /user/balance` (see
 `https://api-docs.deepseek.com/api/get-user-balance`) using the provider's
 `apiKey`. The balance is **account-wide** (one balance for both
-`deepseek-v4-pro` and `deepseek-v4-flash`), refreshed on startup and then every
+`deepseek-v4-pro` and `deepseek-flash`), refreshed on startup and then every
 5 minutes. The line is hidden when no API key is configured or the request
 fails.
 
@@ -172,7 +172,7 @@ rm -rf "$HOME/.cache/opencode/packages/file:/home/mukminov/src/opencode-peak-sch
 {
   "$schema": "https://opencode.ai/tui.json",
   "plugin": [
-    ["/home/mukminov/src/opencode-peak-scheduler", { "peakModel": "deepseek-v4-flash" }],
+    ["/home/mukminov/src/opencode-peak-scheduler", { "peakModel": "deepseek-flash" }],
     "/home/mukminov/src/opencode-git-footer"
   ]
 }
@@ -193,7 +193,7 @@ Plugin options (`plugin` tuple):
 | Key               | Type     | Default                  | Meaning                                  |
 | ----------------- | -------- | ------------------------ | ---------------------------------------- |
 | `providerID`      | `string` | `"deepseek"`             | Provider part of the target model        |
-| `peakModel`       | `string` | `"deepseek-v4-flash"`    | Model used during peak hours             |
+| `peakModel`       | `string` | `"deepseek-flash"`    | Model used during peak hours             |
 | `offModel`        | `string` | `"deepseek-v4-pro"`      | Model used off-peak                      |
 | `timezone`        | `string` | `"Asia/Shanghai"`        | IANA timezone the schedule runs in       |
 | `autoID`          | `string` | `"auto"`                 | Pseudo-model id routed by the window     |
@@ -203,7 +203,7 @@ Provider `scheduler` options (inside `provider.deepseek.options.scheduler`):
 
 | Key         | Type     | Default               | Meaning                              |
 | ----------- | -------- | --------------------- | ------------------------------------ |
-| `peakModel` | `string` | `"deepseek-v4-flash"` | Model used during peak hours         |
+| `peakModel` | `string` | `"deepseek-flash"` | Model used during peak hours         |
 | `offModel`  | `string` | `"deepseek-v4-pro"`   | Model used off-peak                  |
 | `timezone`  | `string` | `"Asia/Shanghai"`     | IANA timezone the schedule runs in   |
 | `autoID`    | `string` | `"auto"`              | Pseudo-model id routed by the window |
@@ -229,7 +229,7 @@ Notes:
 ## Limitations (honest)
 
 - **Explicit model choices are always respected.** Picking a concrete model
-  (e.g. `deepseek-v4-flash` during off-peak) bypasses the scheduler entirely —
+  (e.g. `deepseek-flash` during off-peak) bypasses the scheduler entirely —
   only `auto` is routed by the window, and `input.model` wins per turn.
 - **The provider package must be reachable at that exact `file:///` path.** The
   import is resolved at startup; moving or deleting the source directory while
@@ -250,7 +250,7 @@ Notes:
   provider-level routing for `auto`, but no session-row switching or status UI.
 - **Per-model `options` are not applied to `auto`-routed calls.** OpenCode
   computes provider options from the `auto` model entry (which has none), so
-  `options` on the `deepseek-v4-pro`/`deepseek-v4-flash` entries (e.g.
+  `options` on the `deepseek-v4-pro`/`deepseek-flash` entries (e.g.
   `reasoningEffort`) are bypassed for calls routed via `auto`. Put shared
   options at provider level (`provider.deepseek.options`) or on the `auto`
   entry instead.
@@ -283,8 +283,8 @@ If this plugin saves you money, you can say thanks:
 
 | QR | Address |
 | --- | --- |
-| ![BNB Smart Chain](assets/bsc.png) | **BNB Smart Chain (BEP-20):** `0x1Df93A331CF8D5bE9d382B6d55fe227D6489B2a2` |
-| ![Tron](assets/tron.png) | **Tron (TRC-20):** `TDfatR9JWqcfx5VuTMHDLQP3BN6f8Ynk5r` |
+| ![BNB Smart Chain](https://cdn.jsdelivr.net/npm/opencode-peak-scheduler/assets/bsc.png) | **BNB Smart Chain (BEP-20):** `0x1Df93A331CF8D5bE9d382B6d55fe227D6489B2a2` |
+| ![Tron](https://cdn.jsdelivr.net/npm/opencode-peak-scheduler/assets/tron.png) | **Tron (TRC-20):** `TDfatR9JWqcfx5VuTMHDLQP3BN6f8Ynk5r` |
 
 ## License
 

package/package.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://json.schemastore.org/package.json",
   "name": "opencode-peak-scheduler",
-  "version": "0.1.1",
+  "version": "0.1.3",
   "description": "OpenCode plugin — switch DeepSeek model automatically by Beijing peak/off-peak hours",
   "type": "module",
   "main": "./dist/index.js",