cascade-ai 0.48.0 → 0.49.0

package/README.md

@@ -5,7 +5,7 @@
 
 [![npm](https://img.shields.io/npm/v/cascade-ai?color=aaff00&label=npm)](https://www.npmjs.com/package/cascade-ai)
 [![license](https://img.shields.io/badge/license-MIT-aaff00.svg)](LICENSE)
-[![node](https://img.shields.io/badge/node-%E2%89%A520-5AB4E8.svg)](#installation)
+[![node](https://img.shields.io/badge/node-%E2%89%A522-5AB4E8.svg)](#installation)
 [![providers](https://img.shields.io/badge/providers-6-a78bff.svg)](#ai-providers)
 [![PRs welcome](https://img.shields.io/badge/PRs-welcome-f5a623.svg)](CONTRIBUTING.md)
 
@@ -227,7 +227,7 @@ User prompt
 npm install -g cascade-ai
 ```
 
-> Requires **Node.js ≥ 20**.
+> Requires **Node.js ≥ 22**.
 
 ---
 
@@ -381,11 +381,15 @@ cascade doctor              # confirms detection
     "apiKey":         "...",
     "baseUrl":        "https://YOUR_RESOURCE.openai.azure.com",
     "deploymentName": "gpt-4o",
-    "apiVersion":     "2024-08-01-preview"
+    "apiVersion":     "2024-08-01-preview",
+    "region":         "global"
   }]
 }
 ```
 
+`region` selects the right price sheet — Azure charges ~10% more for `us` and
+`eu` deployments than for `global` ones. Defaults to the global rates.
+
 ### OpenAI-compatible endpoints (Groq, Together, etc.)
 
 ```jsonc
@@ -394,11 +398,28 @@ cascade doctor              # confirms detection
     "type":    "openai-compatible",
     "apiKey":  "...",
     "baseUrl": "https://api.groq.com/openai/v1",
-    "model":   "llama-3.1-70b-versatile"
+    "model":   "llama-3.1-70b-versatile",
+    "local":   false
   }]
 }
 ```
 
+### Is this endpoint free? (`local`)
+
+An OpenAI-compatible endpoint is either your own hardware — llama.cpp, LM
+Studio, vLLM — where inference genuinely costs nothing, or somebody's paid API.
+The same is true of Ollama, which is usually local but can be pointed at a
+rented box. `local` says which:
+
+| `local`   | Meaning                                                            |
+|-----------|--------------------------------------------------------------------|
+| `true`    | Self-hosted. Calls cost **$0**, and Cascade reports $0.            |
+| `false`   | Hosted. Cascade prices calls from its pricing dataset; a model it can't price reports **"cost not tracked"**, never $0.00. |
+| *(unset)* | Inferred: Ollama is local; an OpenAI-compatible endpoint is local when `baseUrl` points at localhost or your LAN. |
+
+Set it explicitly whenever the default guesses wrong — a hosted endpoint
+mistaken for a local one is how real spend gets reported as free.
+
 ---
 
 ## Tools
@@ -797,7 +818,7 @@ web/
 
 | Tool | Required Version |
 |------|-----------------|
-| Node.js | ≥ 20.x |
+| Node.js | ≥ 22.x |
 | npm | ≥ 10.x |
 
 ### Setup