dstack
by dstack · Open-source orchestration for GPU clouds, Kubernetes and on-prem clusters
BenchRank score
Screenshots of dstack
Homepage
Overview
dstack is an open-source control plane for AI workloads that runs them across GPU clouds, Kubernetes clusters and bare-metal or VM hosts reached over SSH. You run the server, add backend credentials in a config file, and manage fleets, dev environments, tasks, services and volumes through a CLI, UI and API. A hosted option, dstack Sky, provides marketplace GPUs.
- Best for
- AI teams running training and inference across mixed GPU clouds, Kubernetes and on-prem hosts.
- Pricing
- No plan prices are published: the core is open source and self-installed, while the Enterprise edition and hosted dstack Sky are handled via demo or contact, with marketplace GPU rates shown from $0.80–1.40/hr for an L40S 48GB up to $6.00–12.00/hr for a B300 288GB.
- Runs on
- Self-hostedCLI
Strengths and trade-offs
Strengths
- Open source; runs in your own cloud accounts or on your hardware
- One control plane for fleets, dev envs, tasks, services, volumes
- Works with NVIDIA, AMD, Tenstorrent and TPU accelerators
- Hosted dstack Sky marketplace if you have no cloud accounts
Trade-offs
- No prices published for the Enterprise edition or dstack Sky
- SSO and air-gapped setup are Enterprise-only, not open source
- Built for containerised workloads only
- You install and run the dstack server yourself
How dstack markets itself
A structured read of the promise, proof and page design on dstack’s captured homepage.
Homepage capture
“The orchestration stack for heterogeneous AI compute”
- Angle: Developer-first
- Hero: 3D render
Pricing
Published plans and prices from dstack’s own pricing page.
How this score is made up
Each dimension is scored out of 100 and combined into the headline score using fixed weights.
MCP support
Whether an agent can drive the product through the Model Context Protocol, and how much setup that takes.
API quality
Public API surface: machine-readable spec, official SDKs, documented auth, errors, rate limits and versioning.
Documentation
Publicly reachable docs — coverage, freshness, code samples and machine readability.
Agent friendliness
How readable the site is to an automated client: llms.txt, structured data, server-rendered content, crawler access.
Changelog
A public, dated record of what shipped and when — the clearest signal that a product is still alive.
Marketing site structure
Whether the site answers a buyer's questions: clear positioning, the pages that matter, and accessibility.
Page speed
How fast the site loads for real visitors: Chrome UX Report 75th-percentile LCP, INP and CLS, with a Lighthouse mobile run standing in where a site has too little traffic for field data.
Operational trust
Status page and incident history, security disclosure, compliance and data-processing documentation.
Measured, but not part of the score
Useful to know, but not a mark for or against the product — so these do not affect the ranking.
Openness
Source availability, self-hosting, data export and open standards. Scored and shown, but not part of the composite — paid SaaS is not worse for being paid SaaS.
Maintenance
Release cadence and repository activity. Scored and shown, but not part of the composite — it is only measurable for open repositories.
This doesn’t look right — report a problem with dstack’s score
Where this comes from
The dstack pages BenchRank reads when it scores the product — its documentation, release notes, status and security pages, and its repository where there is one.
Alternatives in Orchestration & Scheduling
Ranked 1
82.4 — BenchRank score out of 100Prefect
Prefect · Python-decorator workflow orchestration for data, ML and agents
Best for: Python data and ML teams that want scheduled, self-retrying pipelines without hosting a scheduler
Ranked 2
80.9 — BenchRank score out of 100Temporal
Temporal · Durable execution for workflows, AI agents and long-running jobs
Best for: Engineering teams running long workflows, pipelines or AI agents that must survive crashes and retries
Ranked 3
78.9 — BenchRank score out of 100Inngest
Inngest · Durable execution for background jobs, workflows and AI agents
Best for: Teams adding retries, flow control and step-level tracing to background jobs and AI agents.
Is this your product?
Claim dstack to manage its profile. Claiming lets you suggest edits to the descriptive fields — it never changes scores or rankings.