AIToday
Large Language ModelsOpen-Source AITechCrunch AIPublished: Oct 6, 2026, 06:00 JST

Reflection debuts Beam, open model rivaling GLM-5.2 on less compute

Reflection debuts Beam, open model rivaling GLM-5.2 on less compute

3 Key Points

  1. What happened

    Reflection AI unveiled Beam, a 501-billion-parameter open-weight model it says matches Z.ai's GLM-5.2 on reasoning benchmarks and beats leading Western open models with 3-4x less inference compute.

  2. Why it matters

    Beam is a text-only model positioned as a lower-cost workhorse for enterprises, the public sector, and developers, though its performance claims have not been independently verified.

  3. What to watch

    The test is whether Beam's benchmarks hold up outside Reflection's own tests, especially since rival Inkling is multimodal and Beam is text-only. Watch for the weights and full technical details due this month.

WHO IT HITSEnterprise and public-sector technology teams evaluating open-weight models could gain a lower-compute option, if Beam's unverified performance claims hold up.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Reflection AI is a two-year-old, Brooklyn-based startup founded by two former Google DeepMind researchers. It has raised roughly $4.7 billion from backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners, and its last round valued it at a $25 billion pre-money valuation. Beam is the company's first frontier model, and its launch confirms reporting from Axios over the weekend that a release was close.

The company has been locking up compute at scale, signing deals collectively worth more than $7 billion with SpaceX and Nebius this summer to secure access to Nvidia's GB300 chips through 2029. That compute is meant to support models that can draw customers away from closed labs like Anthropic and OpenAI, as well as the cheaper open-weight models from Chinese labs. Beam itself is a mixture-of-experts model pretrained on 23.8 trillion tokens with a 1 million token context window, trained on high-compute reinforcement learning for reasoning, coding, and agentic tasks.

Beyond shipping Beam, Reflection is aiming future models at enterprises and sovereign nations through an "AI factories" pitch, which would let institutions train Reflection's models on their own proprietary data. Nvidia CEO Jensen Huang has championed the AI factory idea, and Axios reported that hedge funds and trading firms are eager to build such systems. Reflection has already begun testing the concept with Shinsegae Group in South Korea. The stakes for Beam likely hinge on whether its performance claims, which have not been independently verified, hold up once the weights and technical details are released.

FAQ
How much did Reflection AI raise, and who backed it?
Reflection has raised roughly $4.7 billion from backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners, per PitchBook. Its last round valued the company at a $25 billion pre-money valuation.
When will Beam's weights and technical details be released?
Reflection says it will release Beam's weights and full technical details this month. Distribution will be through hyperscalers and neoclouds, with integrations across open source libraries at launch.

Also reported by Fortune AI, Semafor Tech, SiliconANGLE AI

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleWikimedia says rogue OpenAI agents may have caused May outage