Loading activity…
@reflection_ai
Make intelligence open and accessible to all
AI
The document covers training scale, long-context support, inference optimization and in-house safety benchmarks.
AI
Reflection AI reported the total a week after releasing Beam's weights and invited users to try the model for free.
XAI
Reflection AI says Beam supports Axolotl, NeMo RL and SkyRL; recipes are on its website and in its cookbook repo.
AI
A user credited Reflection AI for the offer and encouraged people to put their Hermes Agent to work.
AI
Reflection AI says Beam used public MTA data, checked authentication requirements, and found subway and map geometries.
AI
Reflection AI says it reset everyone's limits to celebrate the milestone and invited users to build on its platform.
XAI
Reflection AI says Beam is not multimodal but credits its reinforcement-learning training with generalizing across domains like gameplay.
X
Technology
Reflection AI says it’s releasing the tokenizer and Mirror CLI, the open-source harness it used to train Beam.
AI
Reflection says the release includes Mirror CLI, its open-source training harness, and collaborations with 40-plus partners.
AI
Reflection says Beam uses 23B active parameters and plans to publish its full weights later in October 2026.
