dstack

Infrastructure orchestration is becoming an agent skill. When agents run experiments, training, and eval autonomously, they now can provision compute, schedule workloads, and track state, all inside the loop. New post on what this means for platform teams, providers, and the orchestration layer dstack.ai/blog/agentic-o…

Working on GPUs, inference, or training? Come hang out with us in SF on July 23. We’re bringing together engineers from @NVIDIAAI, @ByteDanceOSS, @radixark, @liquidai, @zml_ai, @GraphsignalAI, Cosmic Labs, Kernelize and more for lightning talks and networking. Organized by @CrusoeAI and @lmsysorg / @sgl_project. RSVP: luma.com/rxsn0u0h

dstack 0.20.27 is out 🚀 This release adds experimental support for Slurm. A new backend lets dstack orchestrate both training and inference workloads on your existing Slurm clusters. Same YAML and CLI as for GPU clouds, Kubernetes, and bare-metal fleets. Full release notes 👇 github.com/dstackai/dstac…

Working on a new @dstackai feature I’m genuinely excited about. One command to deploy an inference endpoint on whatever compute you already have access to: GPU clouds, your own data center, bare metal, or Kubernetes. The agent chooses compute and serving stack, prototypes on real machines, works through deployment issues, and can use profiling/perf signals to improve the setup. Then it turns the working recipe into a scalable endpoint. DM me if you want to try an early version.

AI infra builders in SF, we are meeting on July 23. @CrusoeAI, @lmsysorg SGLang, and dstack are hosting a meetup on GPUs, training, inference, and open source AI systems. 12 short technical talks with speakers from @sgl_project, @ByteDanceOSS, @NVIDIA, @radixark, @CrusoeDev, and more to be announced soon. RSVP: luma.com/rxsn0u0h

Crusoe has been doing great work making GPU infrastructure more accessible for AI teams. Now @CrusoeAI has a new guide on using dstack to manage clusters, training, and inference on Crusoe Cloud: docs.crusoecloud.com/third-party-in…

Already using @dstackai? Crusoe Cloud is now a fully documented backend. Overview, quickstart, and cluster provisioning — everything you need to run training and inference on Crusoe with the same YAML you're already writing. → docs.crusoecloud.com/third-party-in… #MLOps #GPU

dstack 0.20.26 is out 🚀 This release makes run submission noticeably faster. Starting a run on an idle instance now takes ~1–2s instead of ~7s. The server reuses pooled SSH connections to instances instead of reopening one per operation. Full release notes 👇 github.com/dstackai/dstac…

New in @dstackai 0.20.26 release: * @awscloud: Capacity Reservations for Dedicated Instances are now supported. * Performance: The release adds several optimizations, including an SSH connection pool, resulting in faster run startup times.

what’s the coolest venue in SF for a 80-100-person engineering meetup?

dstack 0.20.25 is out 🚀 * Multiple replicas of a gateway for high availability in production inference setups * Support for B200 & B300 instances in @awscloud * It's now possible to submit runs to specific instances Also, we've updated dstack's default Docker images to Ubuntu 24.04 github.com/dstackai/dstac…

dstack is now in the official @jarvislabsai docs 🙌 docs.jarvislabs.ai/dstack/ Orchestration is becoming the default way GPUs get used. Thank you, @jarvislabsai for the integration! If you're looking for on-demand GPUs, they offer H100, L4, H200, RTX PRO 6000.

dstack 0.20.24 🚀 * You can now use @zeddotdev with dev environments; handy for experiments when you need a GPU * Services support gRPC with prefill and decode workers (both vLLM and SGLang) when @lightseekorg SMG is used as the router * @jarvislabsai backend now supports on-demand RTX PRO 6000 github.com/dstackai/dstac…