Reflection Beam
- Score
- 44.4%
- Price
- Not stated
- API
- No
- Weights
- No
WHAT IT DOES
Beam is the first model from Reflection AI, the Nvidia-backed New York lab founded by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou. Announced October 5, 2026, it is a sparse mixture-of-experts model with 501 billion total parameters and 23 billion active, a 1-million-token context, and text-only input and output. Reflection reports 44.4% on DeepSWE v1.1, 80.1% on Terminal Bench v2.1, 80.9% on SWE-bench Verified and 90.5% on GPQA Diamond, broadly level with Z.ai's GLM-5.2 and behind Qwen3.8 Max and Kimi K3, and claims it reaches GLM-5.2's scores with 3 to 4 times less inference compute. Weights, a technical report and a model card are due later in October under Apache 2.0; until then access is an early-access waitlist.
WHY WE PICKED IT
The US has had few serious open-weight models since Llama stalled, and Beam is a credible one under a fully permissive licence. It does not lead its Chinese peers, and Reflection's own tables say so. It is listed for the efficiency claim and the Apache 2.0 licence, ranked below the open models it trails until the weights are out and independently tested.
QUICK FACTS
STRENGTHS
- ✓501B MoE with only 23B active parameters
- ✓Apache 2.0 licence announced for the weights
- ✓Claims GLM-5.2-level scores at 3 to 4 times less inference compute
- ✓1-million-token effective context
- ✓80.9% SWE-bench Verified and 80.1% Terminal Bench v2.1, per Reflection
LIMITATIONS
- ⚠Not yet generally available; weights still undergoing red-teaming
- ⚠Text only
- ⚠Trails Qwen3.8 Max and Kimi K3 on most of Reflection's own tables
- ⚠No pricing published and no independent benchmark results yet
GETTING STARTED
Join the early-access waitlist at platform.reflection.ai. The weights, technical report and model card are due later in October 2026 under Apache 2.0; plan any self-hosting around that, and compare it on your own tasks against GLM-5.2 and Kimi K3, the open models Reflection benchmarks it against.
BENCHMARKS
OFFICIAL LINKS
PRICING
FEATURED IN GUIDES
CLAUDE.md In Practice: What Actually Earns A Place In The File
The most-starred CLAUDE.md on GitHub has four rules, not twelve, and the error rates everyone quotes...
Open Models: What They Cannot Do
The benchmark gap has largely closed: open models now score within a point of the closed frontier on...
Hardware For Local AI Models: What The Memory Shortage Changed
Capacity decides whether a model runs. Bandwidth decides how fast it talks, and it is the number mos...
How To Run AI Models Locally: Hardware, Quantisation And Engines
Running a frontier open model locally is a memory problem before it is a speed problem, and a bandwi...
Uncensored AI Models: What Abliteration Actually Does To Them
Abliteration removes a model's ability to refuse by deleting one direction from its activations. It ...
EXPLORE ALTERNATIVES
Compare Reflection Beam with 5+ similar llms AI tools.
FREQUENTLY ASKED QUESTIONS
Is Reflection Beam free?
Pricing information for Reflection Beam is not specified.
Does Reflection Beam have an API?
No, Reflection Beam does not appear to offer a public API.
What is Reflection Beam best for?
Reflection Beam is best for Best US Open-Weight Contender. Beam is the first model from Reflection AI, the Nvidia-backed New York lab founded by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou. The US has had few serious open-weight models since Llama stalled, and Beam is a credible one under a fully permissive licence. It does not lead its Chinese peers, and Reflection's own tables say so. It is listed for the efficiency claim and the Apache 2.0 licence, ranked below the open models it trails until the weights are out and independently tested.
What platforms does Reflection Beam support?
Reflection Beam supports api.
Is Reflection Beam open source?
No, Reflection Beam is not open source.
How do I get started with Reflection Beam?
Join the early-access waitlist at platform.reflection.ai. The weights, technical report and model card are due later in October 2026 under Apache 2.0; plan any self-hosting around that, and compare it on your own tasks against GLM-5.2 and Kimi K3, the open models Reflection benchmarks it against.
How do I use Reflection Beam?
Reflection Beam is a large language model for text generation, analysis, and conversation. Use the API for programmatic access. Enter prompts or questions to get responses. It excels at 501b moe with only 23b active parameters.
FEATURED ON CURATEDAI
Work on Reflection Beam? You're hand-reviewed in our directory. Add this badge to your site — it links back to this profile.
