Rubik's AI has launched its Sonus-1 range of language models, marking the first model release of 2025. The Sonus-1 family includes several versions: Mini, Air, Pro, and a Reasoning model. Initial benchmarks for the Sonus-1 Pro Reasoning model show strong performance, achieving scores of 97% on GSM-8k and 90.15% on MMLU. The models are designed to handle a variety of tasks, from coding to complex reasoning, and are currently available through a chat interface, with an API expected to be released soon.
🚀 Rubik’s AI Kicks Off 2025 with Game-Changing Sonus-1 Models Rubik’s AI enters the scene with its Sonus-1 family, featuring models for speed, everyday tasks, and advanced problem-solving. With 97% on GSM-8k and 90.15% on MMLU, these benchmarks rival top players. If the results… https://t.co/kEvbQiBZXC
Rubik AI Release Sonus AI Models, Including Reasoning Rubik AI has revealed its Sonus-1 range of LLMs, combining strong performance with adaptability. These models are built to handle everything from coding to complex reasoning. - Mini offers a rapid and budget-friendly… https://t.co/96bbGiZc9N https://t.co/7noxNJoDgT
One of the first model releases of the new year, and it's a whole family (including a reasoning model): Sonus-1 by @RubiksAI. Chat interface only for now, which includes search and image generation, and an API "will be available soon". Once it is, I can benchmark it properly. https://t.co/4mW7loiOHC