Skip to content
Trending storyDeveloping

Reflection AI reports 1.3 billion sandboxes in its largest RL run for Beam

2 articles1 sourcesince Oct 27Last article

Overview

AISummary of 2 articles

Reflection AI says it ran about 1.3 billion sandboxes in its largest reinforcement learning run for Beam, covering RL training, evaluation and ablations.

The company says the fleet peaked at 883,006 vCPUs, 83% of allocatable capacity, across 23 clusters on Google Cloud and AWS, with evals and other RL experiments running at the same time.

Reflection says it plans to use OpenEnv, which it describes as now part of the Agentic AI Foundation, for future runs, and will bring its learnings to the open ecosystem. A later post from the same account, on the same figures, drops the "@" from OpenEnv's handle, with no change to the numbers.

Written by AI from the articles below · updated Oct 27, 1:43 PM ET

Check the sources:

Latest developmentOct 27, 1:34 PM ET
Reflection AI runs 1.3 billion sandboxes across 23 cloud clusters

Article timeline

The articles in this story. Times are ET.

Oct 27
  1. ReflectionOfficial
    Reflection AI runs 1.3 billion sandboxes across 23 cloud clusters

    AIWe ran ~1.3 billion sandboxes in our largest RL run for Beam - spanning RL training, evaluation, and ablations. With evals and other RL experiments running, the fleet peaked at 883,006 vCPUs (83% of allocatable) across 23 clusters on @googlecloud and @awscloud. We plan to use OpenEnv (now part of @AgenticAIFdn) for future runs, bringing our learnings to the open ecosystem.

    GIF from @reflection_ai's post
  2. ReflectionOfficial
    Reflection AI runs 1.3 billion sandboxes in its largest RL run

    AIWe ran ~1.3 billion sandboxes in our largest RL run for Beam - spanning RL training, evaluation, and ablations. With evals and other RL experiments running, the fleet peaked at 883,006 vCPUs (83% of allocatable) across 23 clusters on @googlecloud and @awscloud. We plan to use @OpenEnv (now part of @AgenticAIFdn) for future runs, bringing our learnings to the open ecosystem.

    GIF from @reflection_ai's post

Heat Trend

Current comparable heat
6
Peak comparable heat
10Oct 27, 1:00 PM ET
Change in past 24 hours
–

The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.