Minor documentation fixes

Author: Murtaza Nsair

docs/getting-started/configuration/ai-providers.md

@@ -302,11 +302,12 @@ After successful connection test:
 ### Important Note on Cost Tracking
 
 **Cost Discrepancies**: Tracked costs in MAESTRO may differ from your provider dashboard, especially with API aggregators like OpenRouter. This happens because:
+
 - Aggregators route to different backend providers with varying actual costs
 - Dynamic routing optimizes for speed/availability, not just price
 - Some providers count tokens differently for billing
 
-Tracked costs are typically 40-60% of dashboard charges. This is not a bug - MAESTRO calculates correctly based on advertised rates. For details and testing tools, see [Cost Tracking Discrepancies](../../troubleshooting/common-issues/ai-models.md#cost-tracking-discrepancies).
+Tracked costs are typically 40-60% of dashboard charges, especially with providers like openrouter. This is not a bug - MAESTRO calculates correctly based on advertised rates. For details and testing tools, see [Cost Tracking Discrepancies](../../troubleshooting/common-issues/ai-models.md#cost-tracking-discrepancies).
 
 ## Common Configurations
 

docs/troubleshooting/common-issues/ai-models.md

@@ -151,6 +151,7 @@ This is a known issue with some API providers' pricing and billing. MAESTRO corr
 #### The Problem
 
 Some API providers, particularly aggregators/routers, have inconsistent billing:
+
 1. **API aggregators route to different providers** - Each backend provider may have different actual costs
 2. **Dynamic routing affects pricing** - Aggregators choose providers based on availability and latency, not just price
 3. **API `usage.cost` field may be unreliable** - Sometimes returns values ~100x lower than actual charges
@@ -169,10 +170,10 @@ Here's actual data from testing with a popular model:
 | 64 | 124 | $0.000106 | $0.000002 | $0.000157 | **1.48x** |
 
 **Key Findings:**
-- Token counts are usually accurate across API and dashboard ✅
-- MAESTRO's pricing calculation is correct based on advertised rates ✅
-- Some providers' API `usage.cost` field may be broken ❌
-- Dashboard charges can be inconsistent with advertised rates ❌
+
+- Token counts are usually accurate across API and dashboard 
+- MAESTRO's pricing calculation is correct based on advertised rates 
+- Dashboard charges can be inconsistent with advertised rates 
 
 #### Testing Your Own Costs
 
@@ -212,6 +213,7 @@ python scripts/test_openrouter_pricing.py --api-key YOUR_API_KEY
 #### Technical Details
 
 The discrepancy may be caused by:
+
 - **Aggregator routing**: Services like OpenRouter route to different backend providers with varying costs
 - **Dynamic provider selection**: Aggregators optimize for availability and latency, not just price
 - **Hidden tokens**: Providers may count system prompts or special tokens not reported in API

docs/troubleshooting/faq.md

@@ -10,7 +10,7 @@
 - Dynamic routing optimizes for speed/availability, not just price  
 - Some providers count hidden tokens not reported in their API
 
-Your tracked costs will typically be 40-60% of actual dashboard charges. This is not a bug in MAESTRO - we calculate correctly based on advertised rates. See [Cost Tracking Discrepancies](common-issues/ai-models.md#cost-tracking-discrepancies) for details and workarounds.
+Your tracked costs may typically be 40-60% of actual dashboard charges, especially with providers like Openrouter, that may route your calls to further providers with differing rates. This is not a bug in MAESTRO - we calculate correctly based on advertised rates. See [Cost Tracking Discrepancies](common-issues/ai-models.md#cost-tracking-discrepancies) for details and workarounds.
 
 ### Q: How can I reduce my API costs?