🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Building around a component that changes every time you run it ain’t easy. Using trials is a way to test for consistency of responses. Try it out in a playground in W&B

The full o1 model is amazing. With reasoning_effort=high, it trounces o1-preview on swebench using my little agent framework. This is pass@1 on a shuffled sample of 100. o1-high: 57/100 resolved, 76m tokens o1preview: 44/100 resolved, 94m tokens

My o1-based AI programming agent is now state of the art on SWE-Bench Verified! It resolves 64.6% of issues. This is the first fully o1-driven agent we know of. And we learned a ton building it.

How do you build an autonomous programming agent that dominates SWE-bench Verified? Shawn Lewis cracked it with an o1-based AI agent solving 64.6% of issues—setting a new state of the art! Here’s how OpenAI’s o1, our W&B Weave toolkit, and relentless iteration made it happen. 👇

Incredible progress in AI software development by Shawn Lewis. Using W&B Weave to improve the performance of AI Programming agent to SOTA! Congrats Shawn!

Managing access in W&B Models just got easier! 🎉 Introducing Team Permission Management in Registry, allowing you to add complete teams to a Registry, and assign roles at the team level. New members automatically inherit permissions—ensuring smooth collaboration.

Today we announced that we are being acquired by CoreWeave, the AI Hyperscaler. 🪄🐝 We could not be prouder or more excited to join forces with this team. Our CEO, Lukas Biewald, wrote a blog post with more details: wandb.ai/wandb/wb-annou…

I’m incredibly proud of everything our team at W&B has accomplished, and excited to keep building with the amazing folks from CoreWeave.

Delighted to see this incredible combo of Weights & Biases and CoreWeave come together! Grateful to Lukas Biewald, Chris Van Pelt (CVP) & Shawn Lewis for having built such an epic company. Thankful to have been part of the journey as an early investor and then as an executive. 🥰🎉

Congrats to Tomide & the Circuit Mind team on this amazing AI product launch allowing anyone to generate🤖Chip designs with Architecture to Schematic & Power Design in 60 seconds or less. Learn more below ⬇️

🔥 Last year we lit up SF with nonstop ideas, demos and hallway aha! moments. We’re so back and this go round is all about launching agentic AI apps from prototype to production. Fully Connected 2025 returns to SF, June 17–18. 2 days, 20+ AI trailblazers & 1K+ builders.

Big news: we’re now officially part of CoreWeave, the #1 AI Cloud Platform. This acquisition marks the start of something much bigger and we’re beyond excited to innovate and scale together. More on what’s next from our founder, Lukas Biewald: wandb.ai/wandb_fc/cw-an…

Today, we announced our acquisition of Weights & Biases. This strategic combination strengthens CoreWeave’s ability to power AI innovation by delivering the leading AI Cloud Platform—purpose-built for performance, scale, and expertise. 💥 hubs.la/Q03l1xmv0 #TogetherWeAreMore

👏Congrats to our Venture Partner Phillip Gurbacki (VP Prod W&B) on the successful acquisition of Weights & Biases by CoreWeave. W&B has become a beloved platform among modern AI devs, making integral contributions to how cutting-edge AI systems are built, trained, and deployed.

Less than a month away from FullyConnected! Expect two full days of hands-on learning, real-world case studies, and deep technical dives. Weights & Biases CoreWeave

Today is an exciting day for CoreWeave! We just announced an agreement to acquire Core Scientific, solidifying our position as the leading AI Cloud Platform—purpose-built to deliver unmatched performance and expertise for the AI era. 🚀 coreweave.com/news/coreweave… #PoweringAI

🔥 OpenAI delays, Moonshot slays! 🔥 While Uncle Sam's still polishing his open-weights model, China just took over the entire LLM scene with Kimi.ai's Kimi K2. And yeah, I put this SOTA XLLM (1T params total, that's an XL language model, even if only 32B are active at

🚨 New Qwen model drop: Qwen3-Coder & Qwen3-2507 are now both live in W&B Inference! (unbeatable prices) Junyang Lin and team really cooked. 🔥 Here's the scoop on these new open SOTA models and how you can get $20 of W&B credits to try them out yourself. 👇

Here it is 🔥 The glorious new open SOTA coding model (and it's non reasoning brother) humming on beautiful H200s provided by CoreWeave and cost pennies! We really went all out for this one, day 0 inference support, huge huge effort from the CoreWeave and Weights & Biases