Update .env

.env

@@ -10,6 +10,7 @@
 # Get your API key from: https://novita.ai
 NOVITA_API_KEY=sk_gaMaeJaUy-qQxms1NIgJuov_RotL_NZXMoQbJlNhS6M
 
+
 # Dedicated endpoint base URL (default for dedicated endpoints)
 NOVITA_BASE_URL=https://api.novita.ai/dedicated/v1/openai
 
@@ -31,19 +32,25 @@ DEEPSEEK_R1_FORCE_REASONING=True
 # Token Allocation Configuration
 # =============================================================================
 # Maximum tokens dedicated for user input (prioritized over context)
-# Recommended: 8000 tokens for large queries
-USER_INPUT_MAX_TOKENS=8000
+# Recommended: 32000 tokens for DeepSeek R1 (128K context window)
+USER_INPUT_MAX_TOKENS=32000
 
 # Maximum tokens for context preparation (includes user input + context)
-# Recommended: 28000 tokens for 32K context window models
-CONTEXT_PREPARATION_BUDGET=28000
+# Recommended: 115000 tokens for DeepSeek R1 (leaves ~13K for output)
+CONTEXT_PREPARATION_BUDGET=115000
 
 # Context pruning threshold (should match context_preparation_budget)
-CONTEXT_PRUNING_THRESHOLD=28000
+CONTEXT_PRUNING_THRESHOLD=115000
 
 # Always prioritize user input over historical context
 PRIORITIZE_USER_INPUT=True
 
+# Model context window (actual limit for your deployed model)
+# Default: 128000 tokens for DeepSeek R1 (128K context window)
+# This is the maximum total tokens (input + output) the model can handle
+# Take full advantage of DeepSeek R1's 128K capability
+NOVITA_MODEL_CONTEXT_WINDOW=128000
+
 # =============================================================================
 # Database Configuration
 # =============================================================================
@@ -161,3 +168,4 @@ CONTEXT_SYNTHESIS_MODEL=Qwen/Qwen2.5-7B-Instruct
 # - Use environment variables in production (not .env files)
 # - Set proper file permissions: chmod 600 .env
 
+