A platform that helps businesses compare, test, and optimize AI models, making it easier to select the best-performing and most cost-effective AI for their needs.
Hey fellow builders! I’m @najwa_assilmi, and together with my co-founder @ramadnsyh , we’re building @intura🐰 —a platform to help you experiment, compare, and monitor LLMs with ease.
Everyone’s building LLMs. Few are testing them like they mean it. We believe the real edge comes after the build—when you’re benchmarking, refining, and shipping with confidence.
We like our tools fun, witty and functional. Inspired by Theseus—the maze-solving mechanical mouse🐭 built by Claude Shannon in 1950—we chose a rabbit as our symbol. For us, it represents the curious, twisty journey of finding the best model setup for your users.
We’re currently in beta and would love your feedback. 🐇💌
Try it out—and let us know what you think. We're all ears!
@kirill_a_belov Hi, thank you for your question! We’re running real tests using live AI APIs like OpenAI and DeepSeek to compare models directly. When we hit the API, we collect data such as response time, input, output, and token usage, then use that to analyze and compare performance across models.
In the current beta, we’re actively building out the live experimentation flow—so more is coming soon!
Congrats on the launch! This looks really useful. I’m curious, does it also support comparing expected vs actual outputs? For example, if I expect a "yes" or "no" answer, can I define that and see how each model performs against it? Would love to try it out!
@fragtex_eth Thanks for the kind words and great question! As part of our upcoming roadmap for robust evaluation, we're adding online and offline methods, including label provision. For example, with offline methods, you'll be able to compare expected vs. actual outputs, like your 'yes/no' example, to assess model performance. We use backtesting and sandbox simulations before live experiments to minimize cold starts and maximize ROI. We're happy to discuss this further and show you how it works!
intura
Hey fellow builders! I’m @najwa_assilmi, and together with my co-founder @ramadnsyh , we’re building @intura🐰 —a platform to help you experiment, compare, and monitor LLMs with ease.
Everyone’s building LLMs. Few are testing them like they mean it. We believe the real edge comes after the build—when you’re benchmarking, refining, and shipping with confidence.
We like our tools fun, witty and functional. Inspired by Theseus—the maze-solving mechanical mouse🐭 built by Claude Shannon in 1950—we chose a rabbit as our symbol. For us, it represents the curious, twisty journey of finding the best model setup for your users.
We’re currently in beta and would love your feedback. 🐇💌
Try it out—and let us know what you think. We're all ears!
With ❤️,
Intura
Hey! Are you using a documentation only (e.g. compare prices, prompts, etc.) or a real tests with AI API's?
intura
@kirill_a_belov Hi, thank you for your question! We’re running real tests using live AI APIs like OpenAI and DeepSeek to compare models directly. When we hit the API, we collect data such as response time, input, output, and token usage, then use that to analyze and compare performance across models.
In the current beta, we’re actively building out the live experimentation flow—so more is coming soon!
Congrats on the launch! This looks really useful. I’m curious, does it also support comparing expected vs actual outputs? For example, if I expect a "yes" or "no" answer, can I define that and see how each model performs against it? Would love to try it out!
intura
@fragtex_eth Thanks for the kind words and great question! As part of our upcoming roadmap for robust evaluation, we're adding online and offline methods, including label provision. For example, with offline methods, you'll be able to compare expected vs. actual outputs, like your 'yes/no' example, to assess model performance. We use backtesting and sandbox simulations before live experiments to minimize cold starts and maximize ROI. We're happy to discuss this further and show you how it works!