Skip to main content
πŸ“– The AI Tool Bible
Flyte preview image
Flyte logo

Flyte

βœ“ Text source-checked Nov 17, 2026

Flyte 2 is a durable AI runtime for running training, serving, and agent workloads with automatic recovery from failures.

Source-checked against the vendor's site Β· rebuilt from the vendor's own pagesProfile updated Link checked How we rate

Freemium(not source-checked)AgentsMulti-model (not source-checked)8.3 / 10

In short

Flyte is an open-source, durable AI runtime by Union.ai that runs training, serving, and agent workloads in pure code with automatic failure recovery.

Flyte is an open-source workflow orchestration platform created and shared by Union.ai. The platform runs workloads for training, serving, and agents while automatically recovering from failures.

Users write workflows in pure Python, TypeScript, Rust, Go, or Mojo without a proprietary DSL or YAML configuration. The vendor says that the runtime supports branching, looping, and fan-out using ordinary control flow, recording input and output for every step so crashes resume rather than restart. Flyte is infrastructure-aware and automatically resubmits tasks with corrected resources after infrastructure failures such as OOM kills, preempted spot nodes, or disappearing GPUs.

Native integrations wrap Streamlit, FastAPI, vLLM, SGLang, and Ollama, allowing users to declare the app, image, resources, and scaling while Flyte handles the serving plumbing. The platform integrates with Apache Spark, Ray, PyTorch Elastic, BigQuery, and Snowflake to make existing stacks durable.

Spotify uses Flyte-orchestrated pipelines for complex financial reports. Wayve uses Flyte for scalable orchestration of petabyte-scale vision workloads in autonomous driving R&D.

Union.ai offers an enterprise version of the Flyte platform for orchestrating, shipping, and scaling AI systems on managed infrastructure.

From the vendor's pages

According to Flyte: Flyte is an open-source workflow orchestration platform created and shared by Union.ai. Its pages list this strength: Workflows are written in plain async functions without a proprietary DSL or YAML configuration.

β€” The AI Tool Bible editorial team

Pros

  • βœ… Workflows are written in plain async functions without a proprietary DSL or YAML configuration.
  • βœ… The system automatically recovers from infrastructure failures like OOM kills and disappearing GPUs by resubmitting with corrected resources.
  • βœ… The platform retries and resumes from the failed task with the exact same inputs, avoiding the need to rerun the entire pipeline.
  • βœ… The vendor says that infrastructure scales up and down to match workload demand, right-sizing every task to reduce idle GPU hours.
  • βœ… Native integrations handle the serving plumbing for Streamlit, FastAPI, vLLM, SGLang, and Ollama applications.

Facts

Platforms
api (not source-checked)
Pricing
freemium (not source-checked)
API available
Yes (not source-checked)
Free trial
Yes (not source-checked)
Open source
Yes
Model used
Multi-model (not source-checked)
Hacker News
10 stories Β· top 145 points Β· last Mar 5, 2026 Β· checked Nov 11, 2026
Earliest Wayback capture
2013-02-14 (earliest archived capture of the site, not a launch date)

Where this tool sits

Use cases

AI trainingModel servingAgent orchestrationBatch inferenceFine-tuning

Frequently asked

How much does Flyte cost?
Flyte operates on a freemium model. While specific pricing details are not documented, Union.ai offers an enterprise version for orchestrating AI systems on managed infrastructure.
What programming languages does Flyte support?
Users can write workflows in pure Python, TypeScript, Rust, Go, or Mojo. The platform does not require a proprietary DSL or YAML configuration, using ordinary control flow instead.
How does Flyte handle infrastructure failures?
Flyte is infrastructure-aware and automatically resubmits tasks with corrected resources after failures like OOM kills, preempted spot nodes, or disappearing GPUs. It records inputs and outputs so crashes resume rather than restart.
Which AI frameworks and tools integrate with Flyte?
Flyte has native integrations for Streamlit, FastAPI, vLLM, SGLang, and Ollama. It also integrates with Apache Spark, Ray, PyTorch Elastic, BigQuery, and Snowflake to make existing stacks durable.
Who uses Flyte in production?
Spotify uses Flyte-orchestrated pipelines for complex financial reports. Wayve uses it for scalable orchestration of petabyte-scale vision workloads in autonomous driving R&D.

Compare Flyte with similar tools

All alternatives β†’

Explore related

Reviews