Update README: streamline benchmarks and quickstarts

README.md

@@ -69,29 +69,26 @@ model-index:
 
 ## ⚡ Executive Summary
 
-Modern AI agent frameworks (Antigravity, Claude Code, Cursor, LangChain) spend up to 40% of their compute budget and hundreds of milliseconds waiting for frontier LLMs (Claude 3.5 Sonnet, GPT-4o) to decide which tool or execution tier to invoke.
+Modern AI agent frameworks (Antigravity, Claude Code, Cursor, LangChain) spend substantial compute budgets and latency classifying routing intents, triaging tool dispatches, or determining next actions.
 
 **`HomeRouter-S1`** is a non-autoregressive, calibrated Small Decision Model (SDM) fine-tuned on 20,582 multi-turn agent execution turns. Built upon ModernBERT and the `LayaForDecision` architecture, it classifies agent tool intents and tier triage in a single forward pass without autoregressive text decoding:
 
 * **Sub-26ms CPU Latency (ONNX)** / **Sub-43ms Apple Silicon Latency (Ollama Metal)**.
-* **Zero Output Tokens ($N_{\text{out}} = 0$)**: Bypasses generative LLM decode loops entirely.
+* **Zero Output Tokens ($N_{\text{out}} = 0$)**: Evaluates entire decision candidate spaces in a single forward pass, bypassing generative decode loops entirely.
 * **100% Strict Enum Guarantee**: 0.00% schema parse errors, eliminating JSON formatting repair loops.
-* **Calibrated Confidence Gating**: Fitted temperature ($T = 1.0618$) achieves $\text{ECE} \le 0.0363$. Applying a threshold of $\tau = 0.90$ escalates ambiguous prompts to Tier-2 frontier models while safely auto-dispatching high-confidence turns.
-* **Avoided Frontier Spend**: Saves up to **$4,500.00 per 1M turns** in avoided frontier API calls (FOCUS 1.5 normalized).
+* **Calibrated Confidence Gating**: Fitted temperature ($T = 1.0618$) achieves $\text{ECE} \le 0.0363$. Applying a threshold of $\tau = 0.90$ escalates ambiguous prompts to secondary tiers while safely auto-dispatching high-confidence turns.
+* **Local Silicon Efficiency**: Eliminates external network roundtrips and recurring per-token cloud inference costs.
 
 ---
 
 ## 📊 Dual-Engine Silicon Benchmark
 
-Tested across local host CPU and Apple Silicon Metal against frontier cloud models:
+Performance telemetry across local host CPU and Apple Silicon Metal runtimes:
 
-| Model / Architecture | Mode / Runtime | TTFD (P50) | TTFD (P95) | Output Tokens ($N_{\text{out}}$) | Parse Error Rate | Calibrated ECE | FOCUS 1.5 Avoided Spend / 1M Turns |
-| :--- | :--- | :--- | :--- | :--- | :--- | :--- | :--- |
-| **Claude 3.5 Sonnet** | Cloud API (Prompt JSON) | 1,420 ms | 3,180 ms | ~280 tokens | 2.14% | 0.0820 | Baseline ($0.00) |
-| **GPT-4o** | Cloud API (Structured Output) | 960 ms | 2,410 ms | ~210 tokens | 1.80% | 0.0760 | -$700.00 |
-| **DeepSeek R1** | Cloud API (Reasoning Loop) | 4,200 ms | 8,900 ms | ~1,200 tokens | 0.95% | 0.0910 | +$1,700.00 |
-| **Homestead Native ONNX (CPU)** | **Direct FP32 Forward Pass** | **25.69 ms** | **26.00 ms** | **0 tokens** | **0.00%** | **0.0363** | **$4,500.00 (100%)** |
-| **Ollama MLX (Metal)** | **Native LayaForDecision** | **40.97 ms** | **42.50 ms** | **0 tokens** | **0.00%** | **0.0363** | **$4,500.00 (100%)** |
+| Model / Architecture | Mode / Runtime | TTFD (P50) | TTFD (P95) | Output Tokens ($N_{\text{out}}$) | Parse Error Rate | Calibrated ECE |
+| :--- | :--- | :--- | :--- | :--- | :--- | :--- |
+| **Homestead Native ONNX (CPU)** | **Direct FP32 Forward Pass** | **25.69 ms** | **26.00 ms** | **0 tokens** | **0.00%** | **0.0363** |
+| **Ollama MLX (Metal)** | **Native LayaForDecision** | **40.97 ms** | **42.50 ms** | **0 tokens** | **0.00%** | **0.0363** |
 
 ---
 
@@ -168,22 +165,6 @@ ollama create olanigan/HomeRouter-S1 -f ollama_laya/Modelfile
 ollama run olanigan/HomeRouter-S1 "find all *.ts files matching client"
 ```
 
-### 4. TypeScript / Node via `s1sdk`
-```typescript
-import { S1Client } from "@homestead/s1sdk";
-
-const s1 = new S1Client();
-const route = await s1.route("Update the database migration script", [
-  "edit_file",
-  "write",
-  "run_command",
-  "tier_frontier"
-]);
-
-console.log(route.choice);      // "edit_file"
-console.log(route.confidence);  // 0.9412
-```
-
 ---
 
 ## 📦 Package Contents & File Sizes