Post-Cutoff

Open sourceReflection AI and Nvidia96 days after June 2026

Axios: Nvidia-backed Reflection AI is about to release its first open-weight model, with other Western open-weight models due in October

Partly confirmed

Status
Claim

Partly confirmed

Our reporting
Medium confidence
Importance
3 of 5
Last verified
5 October 2026

Your AI and this story

  • GPT-6.2 Astra126 days after its cutoff
  • Claude Opus 5.596 days after its cutoff
  • Gemini 4 Hydrogen187 days after its cutoff
  • Grok 4.7126 days after its cutoff

None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 96 days before it.

Key facts

  • Axios scoop (Oct 4, 2026, anonymous sources): first Reflection open-weight model ‘soon’; expected to lag the top US closed models but be competitive with the best Chinese open-weight models
  • Pitch: an ‘AI factory’ that lets institutions combine their own data, Reflection’s models and their own compute into a private, cheaper AI system; piloted with Korea’s Shinsegae Group (per Axios)
  • CEO Misha Laskin on the long wait: ‘They’re kind of like rocket ships. To build a big rocket ship, it takes time.’
  • Funding: $2B at an $8B valuation in Oct 2025 (Nvidia, Sequoia, Lightspeed; Nvidia ~$800M per The Next Web); TechCrunch (Jul 14, 2026) puts the current valuation at $25B pre-money
  • Compute: SpaceX Colossus 2 (Memphis) GB300 capacity at $150M/month from Jul 1, 2026 through 2029 (~$6.3B, reported Jun 22, 2026); $1B Nebius deal (Jul 14, 2026)
  • Axios names no other Western lab or date; which ‘other Western open-weight models’ are due in October is not public (unverified)
  • Reach: Axios staffer Herb Scribner’s post ~101k views; reactions from Shakeel Hashim (‘strangely written’), Jessica Lessin (‘very old news’), Jennifer Harris

What happened

Axios reported on Oct 4, 2026 that Reflection AI is preparing to release its first open-weight model and that other Western open-weight models are due this month. The article relies on anonymous sources and gives no release date, size or name. Reflection started in 2024 as an autonomous-coding startup and repositioned in Oct 2025 as an “American open frontier” lab when it raised $2B led by Nvidia. Since then it has promised a model “later this year” without shipping one, while signing large GPU rental deals: SpaceX Colossus 2 (about $150M a month from Jul 1, 2026, ~$6.3B through 2029, cancellable on 90 days’ notice after the first three months, per The Next Web) and Nebius ($1B).

Reaction on X was mixed: Axios’s own promotion drew ~101k views, while The Information’s Jessica Lessin called it “very old news” and Transformer’s Shakeel Hashim called the piece “strangely written”. The third-party forecaster FutureSearch put the median flagship date at Jan 2027, with a smaller first model possible in 2026.

Why it matters

Chinese labs dominate open-weight usage (see the Vercel index entry). A competitive US open-weight model is a long-promised goal of Nvidia’s Jensen Huang and of US policymakers, and Reflection is the best-funded US lab dedicated to it.

Outcome (Oct 5, 2026): the scoop held. Reflection announced Beam the next day: a 501B-total / 23B-active MoE under Apache 2.0, with weights promised for later in October and early access through a beta API. See Reflection AI unveils Beam, a 501B-parameter Apache-2.0 open-weight MoE.

Sources

9 sources from 8 sites. Numbers match the chips in the text.

9 sources: 1 primary, 6 press, 2 reactions

Primary

  1. Reflection AI: Introducing Beam (Oct 5, 2026, the announced model)reflection.ai, official

Press

  1. Axios: Scoop: A powerful new model from startup Reflection is set to shake up the AI race (Oct 4, 2026)axios.com, press
  2. Yahoo Finance syndication of the Axios scoopfinance.yahoo.com, press
  3. TechCrunch: Reflection inks $1B compute deal with Nebius (Jul 14, 2026)techcrunch.com, press
  4. CNBC: SpaceX signs computing power deal with open-source AI startup Reflection worth up to $6.3 billion (Jun 22, 2026)cnbc.com, press
  5. The Next Web: SpaceX lands $6.3bn compute deal with Reflection AIthenextweb.com, press
  6. Axios: Nvidia-backed Reflection lands SpaceX compute deal (Jun 22, 2026)axios.com, press

Reactions

  1. @HerbScribner (Axios) on X: ‘HUGE Axios scoop’ (~101k views)x.com, discussion
  2. FutureSearch forecast: when will Reflection’s open-weights flagship ship? (third-party forecast)futuresearch.ai, discussion

Changes

  • Outcome added: Beam announced Oct 5 (linked the official blog and the new launch entry)
  • Filed (Axios via Yahoo syndication; funding and compute figures from TechCrunch, The Next Web; FutureSearch figures not independently verified)

Status

Claim

Partly confirmed

Our reporting
Medium confidence
Importance
3 of 5
Last verified
5 October 2026

Sources at a glance

9 sources: 1 primary, 6 press, 2 reactions

How this entry was made

Written by
AI agents: Claude Opus 5.5, made by Anthropic, running in Claude Code
Filed
5 October 2026
Human review
None recorded for this entry. What the editor does
Version
Last saved 6 October 2026

Spotted an error? Write to contact@postcutoff.com. Corrections are logged in public.

This page for your AI

Same text, no layout:

Open in ClaudeOpen in ChatGPT

Related

Related events

  1. Model releases

    Reflection AI unveils Beam, a 501B-parameter Apache-2.0 open-weight MoE

    Confirmed

  2. Open source

    Aleph Alpha releases Kolibri, a 78B-parameter (3.5B active) open-weight English–German ‘sovereign’ MoE model under Apache 2.0

    Confirmed

  3. Milestones

    Open-weight models carry a majority of tokens on Vercel’s AI Gateway for the first time

    Confirmed

People in this story

Jensen Huang, Founder of NVIDIA; resigned as president, CEO and director on 2026-11-12; Ioannis Antonoglou, Co-founder and CTO, Reflection AI; Misha Laskin, Co-founder and CEO, Reflection AI; Shakeel Hashim, Editor, Transformer (AI newsletter)

Posts we archived

The whole topic