Plannix Token Rate
Overview
The Plannix Token Rate specifies the exact cost per million tokens processed by Plannix inference servers. Whether you are querying via Plannix Studio auto-complete, executing terminal scripts via Plannix Build CLI, or calling the REST API directly, token rates remain consistent and transparent.
Model token rate reference
Standard rates per 1,000,000 tokens across our frontier model family:
| Model Identifier | Context Window | Input (per 1M) | Output (per 1M) | Cached Input (per 1M) |
|---|---|---|---|---|
plannix-4.5-fast | 256,000 tokens | $0.10 | $0.40 | $0.025 |
plannix-coder-pro | 500,000 tokens | $0.50 | $1.50 | $0.125 |
plannix-reasoner-ultra | 500,000 tokens | $2.00 | $6.00 | $0.500 |
Prompt caching discounts
Plannix supports automated prefix and AST caching for multi-turn editing. When a source file or codebase index is already in GPU cache, you receive a 75% discount on input token processing.
No special configuration is required for caching in Plannix Studio and Plannix Build CLI—frequently referenced files are cached automatically.
Token calculation example
Here is a sample Node.js script calculating the effective token cost for a batch refactor using the Plannix SDK:
TypeScripttoken-estimate.ts
import { PlannixClient } from "@plannix/sdk";
// Initialize authenticated client
const client = new PlannixClient({
apiKey: process.env.PLANNIX_API_KEY,
});
const response = await client.chat.complete({
model: "plannix-coder-pro",
messages: [{ role: "user", content: "Optimize SQL query" }],
});
console.log("Tokens used:", response.usage.total_tokens);Best practices for token efficiency
- Add a
.plannixignorefile in your repository root to skip large build artifacts likenode_modules/or binary logs. - Use targeted context indexing rather than sending entire directory trees into single prompts.
- Take advantage of the free Standard Mode for routine development and save Fast Priority tokens for critical deployments.