Timeline
- 7.5
Welcome Beam: a 501B open model you can own, customize, and scale
Reflection introduced Beam, described as its first open-weight model, a sparse Mixture-of-Experts model with 501 billion total parameters and 23 billion active parameters, built for coding, reasoning and agentic workloads. The instruction-tuned model was released on Hugging Face in BF16, FP8 and NVFP4 numerics. Reflection positions Beam as a model users can own, customize and scale, which puts a frontier-scale open-weight option in the hands of builders who want to fine-tune or self-host rather than rely solely on hosted APIs. Beam supports a context window of 1 million tokens and was developed predominantly for English; it takes text as input and generates text as output.
- 7.5
Nvidia in talks to acquire US AI startup Reflection AI
Nvidia is in early-stage talks to acquire US open-weights model developer Reflection AI or to increase its investment in the company, people familiar with the matter said, according to a Financial Times report cited by BlockBeats on October 11. The deal could take several forms, including an acqui-hire, further equity investment, or agreements to supply Reflection with more chips and compute, and the two companies could reach an agreement within weeks. Nvidia is already one of Reflection's largest shareholders, and an acqui-hire structure would let it absorb the startup's staff and license its technology without the lengthy regulatory review a full acquisition typically requires. The talks are still at an early stage and no agreement has been reached. Reflection AI was valued at $25 billion in a funding round in March.
- 7.0
Reflection debuts Beam, an open-weight AI model to rival Chinese models at lower compute cost
Reflection AI unveiled Beam on Monday, October 5, 2026, its first frontier open-weight model. According to the company's blog, Beam is a sparse Mixture-of-Experts model with 501 billion total parameters and 23 billion active, built for coding, reasoning and agentic workloads. The launch puts a US lab in direct competition with lower-cost Chinese open-weight models such as DeepSeek and Moonshot AI's Kimi in coding and related tasks, a segment where Chinese releases have set the cost benchmark. Reflection is Nvidia-backed and is pitching Beam and future models at enterprises and sovereign nations through an "AI factories" offering that would let institutions train the models on their own proprietary data. The available coverage does not include independent benchmark results or licensing terms.