Flyte
Flyte is a durable AI runtime for workloads involving training, serving, and agents.
What Flyte is
Flyte is a durable AI runtime for workloads involving training, serving, and agents. Workflows are authored as plain Python async functions, with a TaskEnvironment for specifying images, resources, secrets, and retry policy. Flyte supports dynamic execution through branching, loops, and runtime decisions. If a task fails, it can resume from that task and recover from infrastructure failures such as OOM kills, preempted nodes, or disappearing GPUs. For serving, native integrations include Streamlit, FastAPI, vLLM, SGLang, and Ollama, with replicas able to scale to zero when idle. Flyte can run locally with a devbox or on a user's cluster, and its listed integrations span tools such as Apache Spark, Ray, BigQuery, Snowflake, and MLflow. The Flyte OSS runtime is free and open source under Apache 2.0. Union.ai's Team plan is $950 a month, with usage charges and a 30-day free trial; Enterprise pricing is custom.
Good
- Free Apache 2.0 open-source runtime
- Resumes failed tasks and recovers from infrastructure failures
- Supports branching, loops, and runtime decisions
- Serving integrations include Streamlit and FastAPI
- Can scale serving replicas to zero when idle
Watch out
- Team plan has a $950.00 USD monthly minimum
- Team plan limits runs to 25,000 actions
- Team plan includes one cluster
Who it suits
Flyte suits teams building AI workflows for training, serving, or agents, particularly those needing dynamic execution and recovery from infrastructure failures. It offers local devbox and cluster deployment options.
Verdict
Flyte combines workflow authoring, dynamic execution, failure recovery, and serving support for AI workloads. Its OSS runtime is free, while Union.ai's Team plan adds a monthly minimum and usage charges.
Plans and prices
From Flyte's own pricing page, checked 15 Oct 2026.
| Plan | Price | What it gives you |
|---|---|---|
| Flyte OSS | Free | Self-managed infrastructure · Flyte 2 documentation describes deployment on your own cloud infrastructure · Free and open source; users deploy and manage it themselves |
| Team | $950 a month | 30-day free trial · $950 monthly minimum credited to usage · 1,000 concurrent actions · 30-day data retention · 1 cluster · $950/ mo + usage |
| Union.ai Team (enterprise Flyte platform) | $950 a month | Metered durable actions · vCPU-hours · memory · GPU-hours · AWS Marketplace availability announced September 30, 2026 · Usage-based billing with a $950 monthly minimum credited toward usage; billing starts after the 30-day trial |
| Enterprise | Priced on request | Custom concurrent actions · custom retention · 3+ clusters · white-glove support · Custom |
The facts
Each one read from Flyte's own pages, with the date it was checked.
What it does
- Autoscaling
- Long-running apps can scale replicas up or down automatically and scale to zero when idle.checked 15 Oct 2026
- Batch inference
- Flyte 2 includes a DynamicBatcher for batching inference requests with configurable batch size, queue size and timeout.checked 15 Oct 2026
- Caching and reproducibility
- Executions are automatically logged and versioned, and identical inputs can reuse cached results.checked 15 Oct 2026
- Certifications
- Union.ai's security page lists SOC 2 Type II, SOC 2 Type I, and HIPAA.checked 8 Oct 2026
- CLI
- Flytectl is a Go-based cross-platform command-line client for FlyteAdmin, with Homebrew and shell-script installation instructions.checked 15 Oct 2026
- Deployment
- Flyte 2 can run locally with a devbox or on a user's cluster; the site provides `pip install flyte` and `flyte start devbox` setup commands.checked 8 Oct 2026
- Deployment architecture
- Legacy deployment documentation describes sandbox, single-cluster and multi-cluster Kubernetes deployment paths, with production use requiring a real object store.checked 15 Oct 2026
- Developer requirements
- Legacy development documentation supports Python 3.8 or newer and installs Flytekit with pip.checked 15 Oct 2026
- Documentation depth
- The current documentation includes version selectors for Flyte 2 and Flyte 1 plus guides covering tasks, apps, agents, integrations and CI/CD deployment.checked 15 Oct 2026
- Durable execution
- Flyte 2 can recover from infrastructure failures including OOM kills, preempted nodes and disappearing GPUs.checked 15 Oct 2026
- Dynamic execution
- Flyte supports branching, loops, and runtime decisions for dynamic workflows and agents.checked 8 Oct 2026
- Dynamic workflows
- Workflows use ordinary Python control flow for branching, loops and runtime decisions rather than a separate DSL.checked 15 Oct 2026
- Extensibility
- Flyte integrations are implemented through a plugin system and the project invites users to build and submit their own integrations.checked 15 Oct 2026
- Failure recovery
- Flyte can resume from a failed task and automatically recover from infrastructure failures such as OOM kills, preempted nodes, or disappearing GPUs.checked 8 Oct 2026
- Independent praise: Kubernetes fit
- Coding Protocols praises Flyte's typed pipelines, Kubernetes execution and per-task resource control, recommending it for teams already deep on Kubernetes.checked 15 Oct 2026
- Independent praise: typing and reproducibility
- Reddit users repeatedly praise Flyte's enforced typing, caching, versioning and reproducibility for machine-learning workflows.checked 15 Oct 2026
- Integrations
- The Flyte site lists integrations including Apache Spark, Ray, PyTorch Elastic, BigQuery, Snowflake, Weights & Biases, OpenAI Agents SDK, LangGraph, CrewAI, Pydantic AI, MLflow, Dask, and Databricks.checked 8 Oct 2026
- License
- The Flyte repositories and SDKs are published under the Apache License 2.0.checked 15 Oct 2026
- Limit: ecosystem gaps
- A Reddit reviewer says some features were missing in Flyte 1 and that they still had to store artifacts in MLflow.checked 15 Oct 2026
- Limit: Kubernetes dependency
- Independent comparison guidance positions Flyte primarily for teams already operating Kubernetes, implying a poor fit where that platform expertise is absent.checked 15 Oct 2026
- Limit: operational complexity
- Reddit reviewers report setup headaches involving permissions, configuration, firewalls and cloud deployment.checked 15 Oct 2026
- Logical recovery
- Failed tasks can be retried and resumed with the same inputs instead of rerunning the whole workflow.checked 15 Oct 2026
The company
- Company facts
- LinkedIn lists Union.ai as privately held, founded in 2021, headquartered in Seattle and employing 11–50 people.checked 15 Oct 2026
- Company history
- Union.ai's company page says Flyte was open-sourced in 2020 and Union.ai was founded with $10 million in seed funding in 2021.checked 15 Oct 2026
Alternatives to Flyte
- 1
Windmill
Free planOpen sourceWebAPILinuxSelf-hosted8.3From $120 a month - 2
Kestra
Free planOpen sourceWebAPILinuxmacOS7.9Free plan - 3
ActiveBatch
AndroidiOSWebAPI7.6$10,000 one-time - 4
Apache DolphinScheduler
Free planOpen sourceWebAPILinuxSelf-hosted7.3Free plan - 5
Argo Workflows
Free planWebAPILinuxmacOS7.0Free plan - 6
Dagu
Free planWebAPILinuxmacOS7.0From $50 a month