Update README.md

README.md

@@ -50,7 +50,7 @@ Route prompts to appropriate LLM tiers based on predicted complexity:
 | 2 (Moderate) | Complex reasoning, deep knowledge required | Strong model |
 | 3 (Complex) | Very complex reasoning, niche expertise | Frontier API model |
 
-**Recommended routing strategy:** Group levels **0 and 1** together (fast/cheap tier), level 2 as standard, level 3 as premium. The model achieves **92.5% adjacent accuracy** on my test, meaning it rarely misroutes by more than one tier.
+**Recommended routing strategy:** Group levels **0 and 1** together (fast/cheap tier), level 2 as standard, level 3 as premium. The model achieves **93.0% adjacent accuracy** on my test, meaning it rarely misroutes by more than one tier.
 
 ### Out‑of‑Scope Use