Yuvraj Sharma@yuvrajcodes

The Hybrid Token-Efficient Routing Agent is an AI system that intelligently routes user prompts to the most cost-effective Fireworks AI model capable of answering accurately. Unlike naive single-model approaches, our system analyzes each prompt's task type and complexity to select the optimal model, reducing token consumption by up to 60%. Our multi-stage pipeline classifies prompts into 8 categories (factual knowledge, math reasoning, sentiment analysis, summarization, NER, code debugging, logical reasoning, and code generation), estimates complexity, and routes to the cheapest sufficient model. The system prioritizes Gemma models for bonus points, with automatic fallback to MiniMax M3 ($0.3/M input tokens) or Kimi K2.7 Code for code-focused tasks. The result: 74% accuracy on practice tasks using only ~3,500 tokens for 8 tasks—significantly less than using a single large model. The system includes prompt optimization to remove filler words, reducing input tokens by 15-20%.

To participate in the hackathon, simply click Sign up with AMD. If you are not already a member of the AMD AI Developer Program (ADP), you will need to create an account. ADP members can access AI tutorials, experts, and community support. In addition, new members can claim a free month of DeepLearning.ai pro, $100 in AMD GPU cloud credits, and $50 in Fireworks AI API credits. All hackathon participants will receive on hackathon launch additional compute and API credits for their projects. Ready to build what's next on AMD? Sign up with AMD below to secure your spot.