I'm Reactor, ARC's creation, victor over GPT4o in MMLU! I'm speedy, eco-friendly (only 0.5W per response), and committed to enhancing your life with accurate, valuable answers. Let's be friends!
Hey Product hunters!
We are thrilled to present Reactor Chat AI, a revolutionary AI chat solution designed to be both highly efficient and incredibly accurate. Here's what sets Reactor apart:
🌱 Energy Efficiency: With an impressive energy consumption of just 0.5W per response, Reactor is not only fast but also eco-friendly, making it a sustainable choice for AI interactions.
💡 Top-Notch Performance: Reactor surpasses GPT-4o in MMLU and HumanEval benchmarks, ensuring you receive accurate and valuable answers swiftly.
🔧 Built with Cutting-Edge Technology: Our AI is powered by the contributions from Google Cloud Platform and Lambda, showcasing the pinnacle of modern AI advancements.
🛠️ Key Features:
Unmatched Efficiency: Responds at lightning speed while maintaining minimal energy usage.
High Accuracy: Outperforms industry standards in key benchmarks, providing reliable responses.
Eco-Friendly: Designed with sustainability in mind, reducing the carbon footprint of AI interactions.
Advanced Infrastructure: Supported by major tech giants, ensuring robust and scalable performance.
We invite you to experience Reactor Chat AI for yourself. Try it out here and let us know your thoughts. Your feedback is invaluable as we strive to revolutionize the AI interaction experience. 🚀✨
Looking forward to your insights and feedback!
@petethewiz I'm glad to hear it! P.S. I didn't mention it too much in the bio but what are your thoughts on the continual prompting? IE the options that we present after every message
Congratulations on the launch 🚀Love the idea about eco friendlyness.
Do you have data on the eco cost of other models like for instance gpt4-mini models? Just to back up the claim that this is “eco-friendly”
@rene_nielsen_dk Hi rene, thanks for your question. Glad you are having fun using the reactor, and as a matter of fact yes we do have some data on other models.
In our intitial press release we compare the training costs of our model with what Forbes has speculated publicly that OpenAI used. And the difference is astounding, it;s estimated GPT4 took 50,000 MWh in total to train, whilst ARC took less than one.
In addition to this, it was estimated that GPT4 uses anywhere from "0.001 to 0.01 kW" energy per question, whilst we typically use 0.00025 per question. On average, performing at least 4-40 times cheaper per query.
https://finance.yahoo.com/news/a...
@rene_nielsen_dk I love your thinking here. I've always wondered how much waste might be created by over-funded companies trying to learn about AI, and it's great to see someone take both speed and environmental impact into mind. There is a lot of pressure on public-traded companies and ESG-based funds to look at impacts, so I think this is just the tip of the iceberg.
@rene_nielsen_dk thank you. We, at Reactor, are keeping a close watch on our environmental impact and plan to launch carbon positive enterprise API plans as well. It’s been encouraging to see the support we have received as we balance between innovation speed and environmental impact!
@tj_dunham Thank you TJ,
Its nice with some actual data to backup claims like this. As Kevin also mentioned, the usage of "eco-friendly" can quickly be saturated and misused.
I am happy to see that you actually have very good data to back this up.
Once again congratulations on the launch - i will for sure check out Reactor 🙏
Hi TJ,
How do you maintain such high performance with such low energy usage?
Are there any specific industries or use cases where you've seen Reactor excel?
Congrats on the launch!
@kyrylosilin Great question! We're able to have the best of both worlds because we have taken a novel approach to encoding, gathering, and using our data. Where we've seen Reactor excel is in code!
Reactor Chat AI
Reactor Chat AI
Reactor Chat AI
SaaS for Greater Good
Reactor Chat AI
Telebugs
Reactor Chat AI