Products
Not Diamond is an " LLM router" that automates the process of helping you select the best response model based on your inputs, choosing the right model for the right question and continuously optimizing your LLM usage costs.
Provides a full set of optimized tuning capabilities for "LLM Model Routers", allowing you to custom-tune routing rules.
This is a tool for developers.
He's offering the average user a free 100,000 uses per month of his AI chat interface... No mistake, you can type 100,000 times for free, get 100,000 replies, and use GPT-4o, Claude 3 Opus, Gemini 1.5 Pro, Perplexity, etc... Such big expensive models.
Ensemble:ChatGPT Mirror Station (domestic access to GPT4 series models)
principle
Not Diamond automatically recommends the best AI model for each message and learns in real time based on your feedback.
👍/👎 When you like or tap on a reply, Not Diamond immediately learns if the model is performing well on your tips and uses your feedback to improve future recommendations. To see the actual results, you can try tapping a reply and asking the same question again.
✨ You can also click the flash icon to regenerate this response using a different model to compare how it would answer.
📊 Not Diamond is 100% free to use. However, each LLM response will show you metrics for response latency and cost so you can compare the differences.
⚔️ You can turn on Arena mode at any time to compare models in direct competition.
⚙️ You can select and deselect specific LLMs as options by clicking on the Settings tab.
✏️ You can edit the system prompts to improve Not Diamond's response to your questions.
📄 To learn more about how Not Diamond works or to integrate model routing into your own application, you can click the Code Documentation icon.
That's it! To get started, try sending some messages and watch them get routed to the right model. Don't forget to provide feedback so Not Diamond can personalize the routing to your preferences.
Functional Features
Train your own router
You can get started with Not Diamond's base router in less than five minutes. If you have your own evaluation data, Not Diamond allows you to train a custom router optimized for your use case.
Breathtaking speed
Help you choose the best model in the time it takes to process a tokens.
Intelligently balancing quality and cost
Efficiently utilize faster, cheaper models without sacrificing quality.
Joint Tips Optimization Support
Programming the best cues for each LLM ensures that the correct model and cues are always used. No need for manual adjustments and experimentation.
Chat interface
Select Model
Select the model that will answer the question, check the competitive mode and two will be selected from the models for answer comparison
Compare and contrast answers
Parallel answer, because the choice of competitive mode, so the output answer to the default hidden model, select the answer to show the corresponding model type
utilization limit
Entering three questions and calling two large models to answer the questions each time takes up the quota three times, taking up three hundred thousandths of a percent of the monthly quota.