pi-multikey 1.8.1 → 1.10.0

package/README.md

@@ -29,7 +29,7 @@ pi install /path/to/multikey
 /multikey → Add pool… → Preset: B.AI → paste keys one per line (blank line to finish)
 ```
 
-After picking the preset, the endpoint, compat, and all 4 model definitions are wired up automatically. Models are available directly as `bai/<model-id>`, e.g. `bai/hy3`.
+After picking the preset, the endpoint, compat, and all 3 model definitions are wired up automatically. Models are available directly as `bai/<model-id>`, e.g. `bai/hy3`.
 
 ## Presets
 
@@ -40,7 +40,6 @@ Built-in presets decouple "model settings" from "keys". The data comes from b.ai
 | hy3 | 256K / 128K | text | off · low · high |
 | mimo-v2.5 | 1M / 128K | text+image | off · high (official: low/medium/high behave identically) |
 | qwen3.8-flash | 1M / 131K | text+image | off · low · medium · xhigh |
-| glm-5.3-flash | 1M / 131K | text+image | low · high · max (always thinks, no off) |
 
 > Why `null` must be explicit: pi's `getSupportedThinkingLevels` treats `mapped === null` as unsupported and hides that level, but **omitting** it is treated as supported and the level name is sent to the API verbatim; `xhigh` / `max` additionally require an explicit non-null value to be usable.
 

package/README.zh.md

@@ -27,7 +27,7 @@ pi install /path/to/multikey
 /multikey → Add pool… → Preset: B.AI → 逐行粘贴 key（一行一个，留空结束）
 ```
 
-选 preset 后 endpoint、compat、4 个模型的全部设定自动就位，模型通过
+选 preset 后 endpoint、compat、3 个模型的全部设定自动就位，模型通过
 `bai/<model-id>` 直接可用，例如 `bai/hy3`。
 
 ## Presets
@@ -41,7 +41,6 @@ DeepSeek / Tencent / 小米官方文档，并对每个 thinking 档位做过实
 | hy3 | 256K / 128K | text | off · low · high |
 | mimo-v2.5 | 1M / 128K | text+image | off · high（官方：low/medium/high 行为相同） |
 | qwen3.8-flash | 1M / 131K | text+image | off · low · medium · xhigh |
-| glm-5.3-flash | 1M / 131K | text+image | low · high · max（始终思考，无 off） |
 
 > 为什么必须显式写 `null`：pi 的 `getSupportedThinkingLevels` 把 `mapped === null`
 > 视为不支持并隐藏该档，但**省略**会被当作支持并把档名原样发给 API；

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "pi-multikey",
-	"version": "1.8.1",
+	"version": "1.10.0",
 	"description": "One pi provider backed by many API keys: automatic 429 rotation, per-request key leases for concurrent subagents, and a /multikey management TUI",
 	"keywords": [
 		"pi-package",

package/presets.ts

@@ -57,7 +57,7 @@ export const PRESETS: Preset[] = [
 	{
 		id: "b-ai",
 		name: "B.AI",
-		description: "api.b.ai — Hunyuan Hy3, MiMo V2.5, Qwen3.8, GLM 5.3 (4 models)",
+		description: "api.b.ai — Hunyuan Hy3, MiMo V2.5, Qwen3.8 (3 models)",
 		defaultPoolId: "bai",
 		baseUrl: "https://api.b.ai/v1",
 		api: "openai-completions",
@@ -96,17 +96,6 @@ export const PRESETS: Preset[] = [
 				maxTokens: 131_072,
 				thinkingLevelMap: levels({ off: "none", low: "low", medium: "medium", xhigh: "xhigh" }),
 			},
-			{
-				// docs.b.ai/llmservice/models/glm-5-3-flash — always thinks (off unsupported);
-				// reasoning_effort: low/high/max, default max.
-				id: "glm-5.3-flash",
-				name: "GLM 5.3 Flash",
-				reasoning: true,
-				input: ["text", "image"],
-				contextWindow: 1_000_000,
-				maxTokens: 131_072,
-				thinkingLevelMap: levels({ low: "low", high: "high", max: "max" }),
-			},
 		],
 	},
 	{
@@ -199,7 +188,7 @@ export const PRESETS: Preset[] = [
 		id: "cline-free",
 		name: "Cline Free",
 		description:
-			"api.cline.bot — Cline account free tier: DeepSeek V4 Flash, Longcat 2.0, Laguna S 2.1, GLM 5.2 (daily per-model quota, lineup rotates)",
+			"api.cline.bot — Cline account free tier: DeepSeek V4 Flash, Longcat 2.0, Laguna S 2.1, GLM 5.3, Solar Pro 4, Muse Spark 1.3 (daily per-model quota, lineup rotates)",
 		defaultPoolId: "cline",
 		baseUrl: "https://api.cline.bot/api/v1",
 		api: "openai-completions",
@@ -226,16 +215,39 @@ export const PRESETS: Preset[] = [
 				name: "Laguna S 2.1 (Free)",
 				reasoning: true,
 				input: ["text"],
-				// Poolside doesn't publish the window; safe default, tune if needed.
-				contextWindow: 128_000,
-				maxTokens: 16_384,
+				// Window/output measured from the live free tier (tuned in multikey.json).
+				contextWindow: 262_144,
+				maxTokens: 32_768,
+			},
+			{
+				// z-ai/glm-5.2:free retired from the Cline lineup (id answers "model not found");
+				// succeeded by glm-5.3-flash. Always-on reasoning, no effort tiers; vision input.
+				id: "z-ai/glm-5.3-flash",
+				name: "GLM 5.3 (Free)",
+				reasoning: true,
+				input: ["text", "image"],
+				contextWindow: 1_048_576,
+				maxTokens: 128_000,
 			},
 			{
-				id: "z-ai/glm-5.2:free",
-				name: "GLM 5.2 (Free)",
+				id: "upstage/solar-pro4",
+				name: "Upstage Solar Pro 4",
 				reasoning: true,
 				input: ["text"],
-				contextWindow: 200_000,
+				cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
+				contextWindow: 524_000,
+				maxTokens: 16_384,
+			},
+			{
+				// Same Meta model as OpenCode Zen's muse-spark-1.3-contributor-free, served
+				// through Cline's chat-completions endpoint (no Responses API / session
+				// affinity needed). Always-on reasoning; text + image input.
+				id: "meta/muse-spark-1.3-contributor",
+				name: "Meta Muse Spark 1.3 Contributor",
+				reasoning: true,
+				input: ["text", "image"],
+				cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
+				contextWindow: 1_048_576,
 				maxTokens: 131_072,
 			},
 		],

package/tui.ts

@@ -267,7 +267,9 @@ export async function pickMany(
 				if (v.length === 0) {
 					add(theme.fg("warning", " No matches."));
 				} else {
-					for (let i = 0; i < Math.min(v.length, maxVisible); i++) {
+					const start = Math.max(0, Math.min(cursor - Math.floor(maxVisible / 2), Math.max(0, v.length - maxVisible)));
+					const end = Math.min(v.length, start + maxVisible);
+					for (let i = start; i < end; i++) {
 						const item = v[i]!;
 						const active = i === cursor;
 						const check = selected.has(item.value) ? theme.fg("accent", "✓") : " ";