slash-tokens 1.2.0 → 1.3.0

package/README.md

@@ -1,8 +1,17 @@
 # /slash-tokens
 
+[![CI/CD](https://github.com/Wolfe-Jam/slash-tokens/actions/workflows/test.yml/badge.svg?branch=main)](https://github.com/Wolfe-Jam/slash-tokens/actions/workflows/test.yml)
+[![npm version](https://img.shields.io/npm/v/slash-tokens.svg)](https://www.npmjs.com/package/slash-tokens)
+[![npm downloads](https://img.shields.io/npm/dm/slash-tokens.svg)](https://www.npmjs.com/package/slash-tokens)
+[![bundle size](https://img.shields.io/bundlephobia/minzip/slash-tokens)](https://bundlephobia.com/package/slash-tokens)
+[![license](https://img.shields.io/npm/l/slash-tokens.svg)](./LICENSE)
+[![⭐ Star on GitHub](https://img.shields.io/badge/%E2%AD%90_Star-black?logo=github&logoColor=white)](https://github.com/Wolfe-Jam/slash-tokens)
+
 Token Optimization for Context Engineers.
 For anyone building with LLMs. 4.8 KB WASM. Sub-millisecond. Zero dependencies.
 
+> 🆕 **v1.3 — The Opus 4.7 Edition.** Same-day support for Claude Opus 4.7 with measured token calibration (1.16–1.51x). Plus Gemini proxy fix and benchmark harness for any upstream.
+
 ```bash
 npm install slash-tokens
 ```
@@ -10,7 +19,7 @@ npm install slash-tokens
 ```js
 import { preflight } from 'slash-tokens'
 
-const check = preflight(prompt, 'claude-opus')
+const check = preflight(prompt, 'claude-opus-4.7')
 // tokens: 47,000 | cost: $0.71 | 11 cheaper options | save 99%
 ```
 

package/dist/cli.js

@@ -40,12 +40,22 @@ function writeToMemory(content) {
 
 // src/slash.ts
 var WASM_INPUT_OFFSET2 = 4096;
-function slash(content) {
+var CALIBRATION = {
+  "claude-opus": 1.15,
+  "claude-opus-4.7": 1.5,
+  "claude-sonnet": 1.15,
+  "claude-haiku": 1.15
+};
+function slash(content, model) {
   if (!content)
     return 0;
   const len = writeToMemory(content);
   const instance = getInstance();
-  return instance.exports.estimate_tokens(WASM_INPUT_OFFSET2, len);
+  const raw = instance.exports.estimate_tokens(WASM_INPUT_OFFSET2, len);
+  if (!model)
+    return raw;
+  const factor = CALIBRATION[model] ?? 1;
+  return factor === 1 ? raw : Math.ceil(raw * factor);
 }
 
 // src/patterns.ts

package/dist/models.js

@@ -2,6 +2,7 @@
 export const MODELS = {
     // Anthropic
     'claude-opus': { input: 5.00, output: 25.00, context: 1000000 },
+    'claude-opus-4.7': { input: 5.00, output: 25.00, context: 1000000 },
     'claude-sonnet': { input: 3.00, output: 15.00, context: 1000000 },
     'claude-haiku': { input: 1.00, output: 5.00, context: 200000 },
     // xAI

package/dist/slash.d.ts

@@ -1,8 +1,9 @@
 /**
  * Estimate token count for a string.
  * Sub-millisecond. Zero allocations in WASM.
+ * Optional model parameter applies per-model calibration.
  */
-export declare function slash(content: string): number;
+export declare function slash(content: string, model?: string): number;
 /**
  * Estimate token count from raw bytes.
  */

package/dist/slash.js

@@ -1,15 +1,44 @@
 import { getInstance, writeToMemory } from './wasm.js';
 const WASM_INPUT_OFFSET = 4096;
+/**
+ * Per-model calibration factors.
+ * WASM estimator was tuned for Opus 4.6. Newer tokenizers may drift.
+ * Factor > 1.0 means "model uses more tokens than WASM predicts."
+ * Applied as: estimate = wasm_estimate * factor (rounded up).
+ *
+ * Default 1.0 = no adjustment. Update after running bench/calibrate.ts.
+ */
+/**
+ * Calibration factors derived from bench/calibrate.ts (2026-04-16).
+ * WASM under-reports 4.6 by ~12% median. 4.7 tokenizer uses 1.29x more
+ * tokens than 4.6 median. Combined correction with safety margin:
+ *
+ * 4.6: 1.15 (corrects -12% under-report, adds ~3% safety)
+ * 4.7: 1.50 (corrects -28% under-report, adds ~5% safety)
+ *
+ * Slash must NEVER under-report. Over-reporting is safe (go/no-go only).
+ */
+const CALIBRATION = {
+    'claude-opus': 1.15,
+    'claude-opus-4.7': 1.50,
+    'claude-sonnet': 1.15,
+    'claude-haiku': 1.15,
+};
 /**
  * Estimate token count for a string.
  * Sub-millisecond. Zero allocations in WASM.
+ * Optional model parameter applies per-model calibration.
  */
-export function slash(content) {
+export function slash(content, model) {
     if (!content)
         return 0;
     const len = writeToMemory(content);
     const instance = getInstance();
-    return instance.exports.estimate_tokens(WASM_INPUT_OFFSET, len);
+    const raw = instance.exports.estimate_tokens(WASM_INPUT_OFFSET, len);
+    if (!model)
+        return raw;
+    const factor = CALIBRATION[model] ?? 1.0;
+    return factor === 1.0 ? raw : Math.ceil(raw * factor);
 }
 /**
  * Estimate token count from raw bytes.

package/dist/transact.d.ts

@@ -3,7 +3,7 @@ export interface ReportOptions {
     tokens_estimated: number;
     tokens_saved: number;
     model?: string;
-    action: 'prevented' | 'routed';
+    action: 'prevented' | 'routed' | 'pass';
     cost_saved_usd: number;
 }
 export interface ReportResult {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "slash-tokens",
-  "version": "1.2.0",
+  "version": "1.3.0",
   "description": "Token Optimization for Context Engineers. 4.8 KB WASM. Sub-millisecond. Zero dependencies.",
   "main": "dist/index.js",
   "module": "dist/index.js",