Lightning AI ⚡️

An amazing week at @NVIDIAGTC 🚀 From Lightning AI being featured in Jensen Huang’s keynote, highlighting our role in the inference ecosystem, to the NemoClaw announcement, it’s clear agents are moving to real, always-on systems. Thank you to everyone who stopped by our booth, joined our happy hour, or attended Lightning AI CEO's @williamfalcon's talk. From teams in production to developers just getting started, one thing was clear: multi-cloud compute + data in one place and speed that turns AI into real products. Looking forward to what’s next ⚡

We’re bringing 200+ NYC builders and engineers together on April 4 to build personalized AI agents. Partnering with Paul Vann and the Validia team, the focus is on making systems that are actually useful, personal, and safe. Build and deploy a real agent in a day with Lightning Studios + @openclaw. Apply to join → luma.com/0xjqdkau

Founders, engineers, and infra teams at @NVIDIAGTC, join us today for an arcade happy hour. Food, drinks, classic video games, and builders talking shop. Limited capacity. RSVP to get on the list → lnkd.in/eiMacSHX

@henriquemfrazao Please DM me your username and email address.

@LightningAI any chance I can get priority verification for Lightning Studios?

@ZizuResearch Glad it’s helping 🙏 keep us posted on what you build.

Thanks, @lightningai lightning.ai, for much needed compute.

Most clouds were designed for web apps. AI needs something different. Lightning AI CEO @williamfalcon will talk about why AI requires vertically integrated, AI-native cloud infrastructure at @NVIDIAGTC 2026. Add it to your schedule → go.lightning.ai/4saIg5P

GTC week is here! Our experts are there all week talking with enterprises building world-class AI systems. Stop by booth 1131 to meet the team!

@Cyberhonk26 Open models still need great tooling. Lightning makes it easier to train, customize, and run them, so more teams can actually build with open weights, not just those with massive infra.

@LightningAI With OpenAI and Anthropic closing their models tighter every quarter, does open source ML infrastructure like Lightning become MORE critical or does it just widen the gap between those with compute and those without?

Join Lightning AI’s contributor program and help build open source AI software trusted by 340,000+ devs and leading AI teams. Work on PyTorch Lightning, Fabric, LitGPT, LitServe, LitData, TorchMetrics, Thunder, and the new LitLogger. Join top engineers and researchers ➡️ t.co/z2DUUypD3D

@koushikch7 Please DM us your username and email address.

@LightningAI any chance I can get priority verification for Lightning Studios? #LightningAI

@208Sva87460 Please DM us your username and email address.

@LightningAI any chance I can get priority verification for Lightning Studios?

@Sirojiddin09 Please DM us your username and email address.

@LightningAI any chance I can get priority verification for Lightning Studios?

@DmaDma413587 Please DM us your username and email address.

@LightningAI any chance I can get priority verification for Lightning Studios?

@tensorfiend @jarvislabsai Love seeing this. Keep us posted on how the training run goes ⚡

@jarvislabsai @LightningAI Damnn, CLI makes things easier . Will try it out🙌🏾

Training a Qwen 3 style 35M parameter model on @jarvislabsai . Training looks ok so far. @LightningAI makes everything super simple 🙇🏽‍♂️

@shetozx Please DM us your username and email address.

@LightningAI any chance I can get priority verification for Lightning Studios?

@zoominai Sorry to hear you are unable to log in. Please email support@lightning.ai and include your new username, old username, and the email address used to set up your account.

@LightningAI My username got changed to "paladinknightmaster-ipl1d" by accident, and now I can’t log in. Is there any way to fix this?

@LightningAI any chance I can get priority verification for Lightning Studios?

Start using Nemotron 3 Super 120B with one line of code. @ArtificialAnlys benchmarked Lightning as the fastest @nvidia output speed across providers. Built for reasoning agents and tool use, run it on Lightning to customize, fine-tune, and deploy agentic systems with one API. Start building→ go.lightning.ai/47GTxmb

@ysu_ChatData Good point. Raw decode speed is only part of the story. We’re actively benchmarking time-to-first-token and tool-calling workloads as well since that’s what actually determines agent latency. Will share results soon.

@LightningAI @nvidia @ArtificialAnlys 480 tokens per second is impressive. Do you have numbers for time to first token and performance under tool calling, not just raw decode speed. That is usually where fast inference turns into useful agents.

Lightning delivered the fastest @nvidia Nemotron 3 Super output speed across providers, according to @ArtificialAnlys.⚡ 480 tokens/sec 🤯 If you're building agentic systems, inference speed matters. Run it in minutes ➝ go.lightning.ai/4d3ufT5

@LearnLord18823 Thanks for the shoutout ⚡ Glad it’s working well for you.

If u want a good VPS, try @LightningAI . They are goat

A year ago, wrapping a research model in a usable web app would take weeks to months. @ARoknabadi needed local Boltz-2 integration for RNA design research at @thernabio. He spun up a Lightning Studio, built a @Gradio interface, and deployed a working structure prediction app the same day. Persistent environments + GPUs + Port Viewer made it a live tool, not just a script. From protein sequence → 3D structure in minutes. Full story → go.lightning.ai/3Nvri31