Experiment tracking and hyperparameter sweeps for AI researchers and engineers, by @CoreWeave. Part of CoreWeave Forge.
San Francisco
Joined May 2018
- Tweets8.9K
- Following1.3K
- Followers49.2K
- Likes12.1K
Pinned Tweet
Introducing W&B Skills for coding agents!
Watch us install the skill, point it at a live fine-tuning project with thousands of traces and RL training runs, then query it all from the terminal in seconds.
Also works with @weave_wb to pull in your agent traces and spot failures.
Weights & Biases retweeted
Five weeks ago, Jev was a secret on our San Francisco floor 🥷
@typesafeai was still in stealth when it sponsored CoreWeave Hacks on September 12 and 13. The team brought Jev before it launched and stayed all weekend helping over 60 teams build on it.
Two days later, @CompleteSkeptic announced Jev to his 1,500 followers. That post is now past 40 million views, and he's at 156K followers.
It hasn't slowed down since.
He told the WSJ that Jev had now crossed a trillion tokens a day and was in use by about a quarter of the Fortune 500.
Workday's new Workday Home is powered by it, and Vercel says nothing was ever adopted faster on its AI Gateway.
Proud that they unveiled it with our builders first. This one's going in the CoreWeave Hacks history books.
We knew Beam back when it was just a loss curve 🍼
🔦 Beam, Reflection's first model with open weights, is live today on CoreWeave Serverless Inference.
Call it through an API and leave the GPUs to us.
We are also proud to share that @reflection_ai used @wandb in CoreWeave Forge during Beam's training.
Developers in the private preview have been calling it fast and accurate, and they like that it doesn't ramble.
See what @TheCodingSoup on our team built with private preview access below.
Get started with Beam here: crwv.co/utuxk
New model dropping soon on @CoreWeave Serverless Inference 👀
CoreWeave Agent Lens (@weave_wb) tracing built in.
Weights & Biases retweeted
During @reflection_ai’s private preview of Beam, I built a haunted escape room to put its coding and reasoning to the test.
I created a game called “Code Your Way Out”, and it's one room with three linked puzzles. Beam has to repair a brass orrery by lining its rings up with a constellation, route the light, then program its escape through a sealed hatch.
Beam never sees the room. Each turn, it gets text observations and writes Python; the code runs in a @CoreWeave Sandbox, and the game engine validates every action before the explorer robot moves.
Beam was out in about 30 seconds. Super bullish on this model and really excited to see builders get this model in their hands.
Weights & Biases retweeted
The @NVIDIAAI team shipped Nemotron 3.5 Super VL today, and you can call it now on CoreWeave Serverless Inference.
Send it text and images, build with supported reasoning, tool calling and structured outputs, and trace every call in CoreWeave Agent Lens.
It's a multimodal mixture-of-experts model with 12B active parameters, served in NVFP4.
Pricing per 1M tokens is $0.25 input, $1.50 output, and $0.05 cached input.
Try it here: crwv.co/utr2P
Weights & Biases retweeted
@CoreWeave Agent Lens already supports OTel, natively. All you have to do is configure your endpoint and we'll slurp all your Claude traces into CoreWeave Forge's purpose built UI. You can do this literally right meow.
docs.coreweave.com/products/…
Weights & Biases retweeted
CoreWeave Model Distillation is in preview in CoreWeave Forge.
Start with a teacher that has "the best big model smell that you can get," as @altryne put it on ThursdAI, and let it teach a small student your one task. GLM-5.3 and its Flash variant are the familiar example.
The student is quick at scale, and that one job stops costing frontier prices.
Try it here: crwv.co/uthth
Weights & Biases retweeted
See a @wandb plot, I like
Weights & Biases retweeted
Your agent handled 40,000 conversations overnight. 230 went wrong. No alarm. All 230 reached customers.
That is why CoreWeave Agent Lens exists.
Log your agent's traces and it reads the entire production corpus, not a sample.
It groups the conversations that fail the same way, even when the wording or the tool sequence differs, and links each group to the traces that prove it. crwv.co/utcU7
Marin 535B-A23B is halfway through its hero run, and the heldout loss is doing what the forecast said it would.
The team built that forecast from four smaller runs before the big one started. Since then, better kernels have taken MFU from 21% to 27%.
It's fully open, built by @percyliang and the Marin Community team, and it trains on @CoreWeave with every change logged in wandb.
True builders will get more from this run's change log than from the finished model's benchmarks.
Pull up a seat to the show here: forge.coreweave.com/wandb/ma…
Weights & Biases retweeted
Your agent made a mistake. Can you turn it into a test?
Zubin Aysola (@wandb) turns an ARIA production trace into an eval and benchmarks the fix.
World’s Fair recording: youtu.be/XyV6bSMyq-I
Meet the builders in NYC, Oct 12–14 → ai.engineer/nyc/2026
🤖 Made with AI
ALT AI Engineer recording graphic on a black and gold background. Headline: Your agent made a mistake. Can you turn it into a test? A surreal glass chamber contains tangled amber and magenta signal ribbons that resolve into an orderly cyan grid of cubes, illustrating a production failure becoming a test. Zubin Aysola, Weights & Biases. World's Fair 2026 recording.
Weights & Biases retweeted
Big changes gonna be announced by @wandb and @CoreWeave.
🤝 forge it.
Weights & Biases retweeted
Here we go! @CoreWeave kicks off with Fully Connected 2026!
Researchers, customers and AI devs all packged into Moscone South - With over 30 breakouts, hands on labs, Battle Bots arena tonight!
Also unveiling Forge - Combining @wandb, @marimo_io, @OpenPipeAI into one platform for AI builders!
RT @CoreWeave: ICYMI: CoreWeave Forge is here 🎉
A production trace that never reaches the next training run is a signal you paid for and t…
Weights & Biases retweeted
🪙 Platinum. Three reports in a row.
@SemiAnalysis_ published ClusterMAX™ 3.0 today, and CoreWeave is the only provider rated Platinum in all three editions.
The report: utm.io/us5Bv
Weights & Biases retweeted
our SENPAI autoresearch found a nice 22% drop in training time for us via writing:
- SliceAttention kernel (we're using Transolver atm)
- BatchedMuon
4 hour saving for a full training run, 26h -> 22h, no significant change in error/quality
github.com/wandb/senpai
AI teams are moving from prototypes to production faster than ever, but rapid development can create new blind spots.
Join us this Wednesday at 10:00 AM PT for the latest What's New Wednesdays session, and get valuable insights on vibe-coding, tracing, and deploying your agents.
Reserve your spot here: utm.io/ur8ga
