Reflection AI launches Beam, a cost-effective open-weight AI model to compete with Chinese counterparts

Oct 5, 2026

AI Summary

Reflection AI has introduced Beam, an open-weight AI model designed to compete with leading Chinese models while significantly reducing compute costs. The model, which features 501 billion parameters, aims to serve enterprises and public sectors by enabling customized AI systems using proprietary data.

Reflection AI has unveiled Beam, its first open-weight AI model, claiming it matches the performance of top Chinese models on advanced reasoning benchmarks at lower costs.
Beam is a text-only mixture-of-experts model with 501 billion parameters and 23 billion active parameters, pre-trained on 23.8 trillion tokens.
The model reportedly uses 3-4 times less inference compute compared to competitors, positioning it as a viable option for enterprises and developers.
Reflection AI was founded in 2024 by former Google DeepMind researchers and has raised approximately $4.7 billion from investors, achieving a pre-money valuation of $25 billion.
The company has secured over $7 billion in deals for compute resources, essential for training advanced AI models, and is exploring partnerships for building customized AI systems.
Beam's weights and technical details are expected to be released this month, with distribution planned through hyperscalers and open-source integrations.

open-weight modelnvidiacompute costai technologybeam

Reflection AI launches Beam, a cost-effective open-weight AI model to compete with Chinese counterparts

Related Stories

Sophos reduces cyber-threat investigation time significantly using OpenAI Daybreak

Website Simulates Elizabeth Holmes' Desk with Court Documents from Theranos Trial

Jev, an AI for making quick decisions, has been a viral hit in Silicon Valley. But OpenAI is hot on its heels

Google introduces Gemini AI with task management capabilities for businesses