OptimiserSim

Created by team TeamExplorer on July 10, 2026
AMD Developer CloudGemma
Hybrid Token-Efficient Routing Agent

OptimiserSim is an intelligent model-routing layer designed to automatically select the least expensive Fireworks AI model capable of answering each query accurately. It evaluates query complexity, routes requests to the most cost-efficient model, and maintains quality through continuous accuracy monitoring. The system delivers 40 to 70 percent cost reduction compared to single-model setups while preserving response fidelity. Analytics provide visibility into routing decisions, cost trends, and performance metrics. The repo includes a FastAPI backend, utility modules, and a Dockerfile for containarized deployment.

Category tags:
GithubPresentationDemo
Explore more applications

Lexia

Jeopardy_500

QuillCraft: Build & Test Your Own AI Tutors

QuillCraft is an educational AI platform that empowers students and educators to design, configure, and test their own AI tutoring bots. Users can define personality, rules, and answer style, then validate their bot’s responses in a real chat interface.

QuillBot

Llama 3.1AMD Developer Cloud

StudySprint

A focused exam-planning app that turns subjects, exam dates, confidence, weak topics, and available study hours into a clear, prioritized revision plan.

Mymrabuild

AMD Developer CloudChatGPTAnthropic Claude

Contxt

Contxt is a portable, privacy-first context layer for every AI. It remembers you across Claude, ChatGPT and Gemini and injects your context live, while an on-device Gemma gateway keeps your private data encrypted so it never leaves your device.

No Thanks

AntigravityGithub CopilotHuggingFace HubHuggingFace SpacesWebGPUOpenAIGemma

Clark

Clark is a general-purpose autonomous web agent. Give it a goal in plain language and it opens a real browser, reasons through the task step-by-step, and executes actions.

Clark

GemmaAMD Developer CloudAI/ML APIAssistants API