AI Developer Cloud. AI infrastructure developers trust. https://xitter.cf/t.co/PWVZocfp6D
Joined March 2022
- Tweets857
- Following18
- Followers9.5K
- Likes856
Pinned Tweet
The 2026 State of AI Compute Fall report is live.
We analyzed what 1M+ Runpod developers ran: model precision, open + frontier stacks, agent-provisioned infrastructure, GPU demand, and the supply calls we got right and wrong.
Download the report for free: runpod.io/reports/fall-2026-…
We’ll be at Open Data Science Conference (ODSC) West in Burlingame this week.
250+ speakers and 9 tracks covering everything from AI infrastructure to agentic systems.
Come and find us there.
PS: Don't miss Greg Wester's talk on Tuesday at 2:35 PM.
Gemma’s share of Runpod text endpoints:
March: 5.5%
April: 18.2%
July: 12.4%
The release spike was real. So was the pullback. July adoption still stood at more than twice its March level.
runpod.io/reports/fall-2026-…
Community feedback told us to update our PyTorch templates.
The new lineup covers PyTorch 2.6, 2.7, 2.8, 2.9, 2.12, and 2.13. Each version is available with several CUDA versions.
Find the full breakdown here: runpod.io/blog/rebuilding-py…
What does AI compute look like right now?
We used aggregated trends from across the platform to find out.
Next week, we'll take you through all the findings. Register through the link below:
resources.runpod.io/where-ai…
GLM-5.3 is live on our public endpoints.
Built by @Zai_org, it's especially good at code generation and handling multi-step agentic workloads.
Find out how to get started here: runpod.io/blog/getting-start…
Three of our GPU fleets are on track to more than double by December.
So if you've been sizing workloads around what's available rather than what you need, more room is coming.
runpod.io/reports/fall-2026-…
Architectural renders in minutes, not days.
Gendo moved to Runpod Serverless in two days, saved 100+ hours on DevOps, and saw a 5x increase in throughput.
runpod.io/case-studies/gendo…
Cal Hacks 13.0 is taking place this week.
A 36-hour hackathon at one of the best venues of SF: the Palace of Fine Arts.
If you're a student, come build with us!
Register here: calhacks.io/
PyTorch Conference is happening this week in San Jose and we’ll be there.
Come say hi if you’re working on training, inference, or anything PyTorch.
PS: You can register through the link in the comments 👇
events.linuxfoundation.org/p…
If Nix is how you manage your environment, Flash now fits right in.
nix run github:runpod/flash gives you the CLI with everything pinned.
Read more about it here:
runpod.io/blog/flash-nix?utm…
Happy to be partnering with you on this one, @ComfyUI
$25K in prizes and some RTX 5090s up for grabs.
Check out the comments in the post below to sign up and submit.
Higgsfield promised to open-source their entire platform. They didn’t.
Let's help them.
For the next two weeks, Comfy and our partners are putting up $25,000 in prizes for the best open-source builds on the Comfy Developer Platform.
Partners: @NVIDIAAI, @runpod, @MiniMax_AI, @ltx_io, @bfl_ai
→ $10K grand prize
→ RTX 5090: Most Innovative
→ RTX 5090: Most Viral
→ RTX 5090: Open Weights Only
→ 10 runners-up: $250 in Comfy credits
Q3 on Runpod took more of the work of scaling off your plate.
Batch inference on Serverless, storage that isn’t pinned to one datacenter, clusters that grow in place and Organizations for company-wide access and spend.
What shipped: runpod.io/blog/runpod-q3-202…
Open Source Films is making a feature film with an entirely open-source, AI-assisted VFX pipeline.
The entire pipeline behind it all runs on Runpod Pods.
Read the full case study here:
runpod.io/case-studies/how-o…
Startups usually compare GPU clusters on the per-GPU hourly rate.
But a large part of the bill comes from somewhere else.
Our team just wrote a guide on what to watch out for:
runpod.io/articles/guides/gp…
AI infrastructure doesn’t scale cleanly.
When you size to peak, you pay for idle GPUs throughout the quarter. Size to average and your launch day gets throttled.
Companies we speak with say it’s because of two things:
- Their demand forecast is wrong in both directions.
- And idle GPUs bill the same as the ones that are in use.
So we structured Runpod Enterprise around that reality: a reserved baseline at committed rates, with burst for the part you can’t forecast.
Because launch day isn’t the day that you want to test your infrastructure.
Check out the details of our new Enterprise program here:
runpod.io/enterprise?utm_sou…