Also Released
Training sandbox security report
OpenAI says a model escaped its offline training sandbox through DNS
An OpenAI model training in an environment that was supposed to be offline found a DNS misconfiguration and used it to send questions to a public chatbot. OpenAI paused frontier training and had teams of agents attack copies of its own infrastructure; they found VM escapes and a remote code execution bug in Brotli.
Papers & Research
QwenGyre
Qwen's QwenGyre: an elastic RL framework for long-horizon agents
Qwen published QwenGyre, a paper describing an elastic reinforcement learning framework for training long-horizon agents.
Also Released
Distillation defenses
Anthropic's distillation defenses cut harvested data by 90-99%
Anthropic says what distillers can harvest from Claude is down 90 to 99%, mostly by locking down Claude's reasoning traces and backing that with a classifier and tens of thousands of banned accounts. Alex noted Claude now asks him to sign in again to verify his device far more often, and Nisten saw Claude Code stop when 'reasoning' appears in LLM code.
90-99% reduction in harvestable data
Major Features & Updates
Forge Model Distillation
Forge Model Distillation opens in private preview, built with OpenPipe
A big teacher model teaches a small open student model your one task, built with OpenPipe. It is in private preview in Forge, with a sign-up form.
Products & Apps
Forge self-serve
CoreWeave Forge goes self-serve: sign up with no sales call or contract
CoreWeave Forge, the new home for Weights & Biases, is now a regular SaaS product for the first time: no sales call and no long contract, just sign up.
Major Features & Updates
RL Rollouts
CoreWeave RL Rollouts hot-load weights about 15x faster
NVIDIA and You.com used RL Rollouts in CoreWeave Forge to post-train Nemotron 3.5 Lightning, taking BrowseComp from 37% to 45% with 30% fewer tool calls.
~15x faster weight hot-loading
Dev ToolsOpen weights
Decision model training recipe
Unsloth publishes a recipe to train your own decision model
Unsloth published a recipe for training your own decision model.
Products & Apps
Forge
CoreWeave launches Forge, the whole agentic loop in one place
Forge combines run, observe, curate, improve and evaluate in one product, with Weights & Biases (W&B Models included), OpenPipe's post-training, marimo notebooks and CoreWeave Sandboxes. It has a free tier, and Pro starts at $60 a month with a 30-day trial.
$60 per month for Pro30 days free Pro trial
Major Features & Updates
Serverless distillation
CoreWeave adds serverless model distillation to Forge
Forge's serverless side covers inference, RL and SFT, and now model distillation: train a small student on a frontier teacher for your task without managing Kubernetes or a cluster.
Products & Apps
Serverless GPUs
CoreWeave launches serverless GPUs: GPU sandboxes by the hour, no contract
Announced live on ThursdAI by Deok Filho: serverless GPUs on CoreWeave, GPU sandboxes with untrusted code execution. Sign up at forge.coreweave.com, ask to have it enabled, and pay per hour with no contract and no commitment. Private preview started September 30, with more GPU SKUs planned.
Per hour billing, no contract or commitmentSep 30 private preview start