DeepSeek: drop /v1 from default baseURL, refresh model list

Provider's published URL is now https://api.deepseek.com (no /v1).
Default model list updated to V4 family (flash + pro). Legacy aliases
deepseek-chat / deepseek-reasoner are still accepted upstream but
deprecate 2026-07-24, so they don't belong in the picker default.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/index.tsx

@@ -66,10 +66,10 @@ export const PROVIDERS: Record<ProviderId, Provider> = {
   deepseek: {
     id: "deepseek",
     label: "DeepSeek",
-    baseURL: "https://api.deepseek.com/v1",
+    baseURL: "https://api.deepseek.com",
     transport: "openai-compatible",
     requiresKey: true,
-    defaultModels: ["deepseek-chat", "deepseek-reasoner"],
+    defaultModels: ["deepseek-v4-flash", "deepseek-v4-pro"],
     hint: "OpenAI-compatible. Create a key at platform.deepseek.com.",
   },
   qwen: {