Reflection AI's Beam, a free-to-download coding AI, is now available on Amazon's cloud
Reflection AI released Beam, an AI built to write and fix software over long jobs. Anyone can download and change it, and Amazon now offers it to run on its cloud.
Reflection AI says Beam is built for coding work that takes many steps. That includes fixing software problems, changing code, using other tools and acting on the results.
The model is "open-weight". That means its core files are published, so developers can download them, run them on their own systems and adjust them. Reflection AI releases them on Hugging Face under the Apache 2.0 license, a permissive open-source license.
Beam is very large. Amazon says it has 501 billion parameters, the numbers a model learns during training, though only 23 billion are used at a time. Reflection AI says it can work with about 1 million tokens of text at once, roughly the size of several long books.
Running it takes serious hardware. Amazon says it needs several powerful GPUs, which are special computer chips for AI. Amazon offers a managed setup in its SageMaker JumpStart service, in two US regions, and a more hands-on option for teams that want more control.
In my opinion, this matters mostly to software teams that want to run a coding AI on their own servers instead of renting one from a big AI company. For most people nothing changes, because Beam needs expensive hardware and is aimed at developers.
The facts
- Reflection AI introduced Beam, an open-weight model for agentic coding and long-horizon tasks such as maintaining code, using tools, interpreting results, and completing multi-step workflows.
- Beam is a sparse mixture-of-experts model with 501 billion total parameters and 23 billion active parameters.
- Reflection reports an effective context length of 1 million tokens.
- Reflection releases Beam's weights through Hugging Face under the Apache 2.0 license, in BF16, FP8, and NVFP4 checkpoint variants.
- At launch, Beam is available in Amazon SageMaker JumpStart in US East (N. Virginia) and US East (Ohio).
- AWS says Beam's 501 billion total parameters require a multi-GPU deployment; the FP8 version runs on ml.p5.48xlarge instances with NVIDIA H100 GPUs, and the NVFP4 version on ml.p6-b200.48xlarge instances with NVIDIA B200 GPUs.
Sources
- AWS ML blog: Introducing Reflection Beam: an open-weight model for agentic coding on AWS
Published
More in New models
New modelsOfficial releaseAyobi’s pickvia Hugging FaceIBM releases AI models that chat, make quick calls and transcribe speech in one
IBM says its new Granite Switch 4.1 models handle conversation, yes-or-no style judgments and speech-to-text from one online connection, using the memory of one model.
New modelsvia The DecoderBlack Forest Labs releases FLUX 3 Image Klein, an image AI for home computers
The new model makes and edits pictures up to 4K size. Black Forest Labs says it runs on an NVIDIA graphics card with 8 GB of memory.
New modelsAyobi’s pickNVIDIA releases open AI models, one built to run on your own PC
NVIDIA says its new Super VL model is five times faster than rivals, and its Bolt model can run on a home PC without the cloud.