- Home
- DataOps Platforms
- Flyte
Flyte
What it is
Flyte 2 is a durable runtime for AI workloads involving training, serving and agents. Users author workflows as plain Python async functions, with a TaskEnvironment for declaring images, resources, secrets and retry policies. Flyte can resume at a failed task and recover from infrastructure problems such as out-of-memory kills, preempted nodes or disappearing GPUs. Dynamic workflows can branch, loop and make decisions at runtime. For serving, native integrations include Streamlit, FastAPI, vLLM, SGLang and Ollama, with replicas able to scale to zero when idle. The listed integrations also include Apache Spark, Ray, PyTorch Elastic, BigQuery, Snowflake, Weights & Biases, OpenAI Agents SDK, LangGraph, CrewAI, Pydantic AI, MLflow, Dask and Databricks. Flyte can run locally with a devbox or on a user's cluster. The free Flyte OSS plan is an Apache 2.0 open-source runtime for local or cluster use. Team costs $950 a month plus usage, with a 30-day free trial and monthly minimum credited to usage. Enterprise pricing is custom and not listed.
For and against
For
- Workflows use Python async functions
- Can resume failed tasks
- Supports branching, loops and runtime decisions
- Can run locally or on a user's cluster
- Free Apache 2.0 open-source runtime
Against
- Team costs $950 a month plus usage
- Team has a 1,000 concurrent action limit
- Team includes one cluster
- Enterprise pricing is custom and not listed
The verdict
Flyte 2 combines workflow authoring, recovery and serving integrations for AI workloads. The OSS runtime is free; hosted Team pricing includes a monthly minimum and usage charges.
Who it suits
Flyte suits developers and teams building AI workflows for training, serving or agents. Its local and cluster deployment options may suit teams that want to run the runtime in their own environment.
Plans and prices
Flyte OSS
Self-managed infrastructure · Flyte 2 documentation describes deployment on your own cloud infrastructure
Read 15 Oct 2026 · union.ai
Union.ai Team (enterprise Flyte platform)
Metered durable actions · vCPU-hours · memory · GPU-hours
Read 15 Oct 2026 · union.ai
Team
30-day free trial · $950 monthly minimum credited to usage · 1,000 concurrent actions · 30-day data retention
Read 8 Oct 2026 · union.ai
Enterprise
Custom concurrent actions · custom retention · 3+ clusters · white-glove support
Read 8 Oct 2026 · union.ai
On the record
- Autoscaling
- Long-running apps can scale replicas up or down automatically and scale to zero when idle.flyte.org · read 15 Oct 2026
- Batch inference
- Flyte 2 includes a DynamicBatcher for batching inference requests with configurable batch size, queue size and timeout.flyte.org · read 15 Oct 2026
- Caching and reproducibility
- Executions are automatically logged and versioned, and identical inputs can reuse cached results.flyte.org · read 15 Oct 2026
- Certifications
- Union.ai's security page lists SOC 2 Type II, SOC 2 Type I, and HIPAA.union.ai · read 8 Oct 2026
- CLI
- Flytectl is a Go-based cross-platform command-line client for FlyteAdmin, with Homebrew and shell-script installation instructions.github.com · read 15 Oct 2026
- Company facts
- LinkedIn lists Union.ai as privately held, founded in 2021, headquartered in Seattle and employing 11–50 people.linkedin.com · read 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.union.ai · read 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.flyte.org · read 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.docs-flyte-legacy.union.ai · read 15 Oct 2026
- Developer requirements
- Legacy development documentation supports Python 3.8 or newer and installs Flytekit with pip.docs-legacy.flyte.org · read 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.union.ai · read 15 Oct 2026
- Durable execution
- Flyte 2 can recover from infrastructure failures including OOM kills, preempted nodes and disappearing GPUs.flyte.org · read 15 Oct 2026
- Dynamic execution
- Flyte supports branching, loops, and runtime decisions for dynamic workflows and agents.flyte.org · read 8 Oct 2026
- Dynamic workflows
- Workflows use ordinary Python control flow for branching, loops and runtime decisions rather than a separate DSL.flyte.org · read 15 Oct 2026
- Extensibility
- Flyte integrations are implemented through a plugin system and the project invites users to build and submit their own integrations.union.ai · read 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.flyte.org · read 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.codingprotocols.com · read 15 Oct 2026
- Independent praise: typing and reproducibility
- Reddit users repeatedly praise Flyte's enforced typing, caching, versioning and reproducibility for machine-learning workflows.reddit.com · read 15 Oct 2026
