What Fly shipped in 2026

Update: May 31, 2025

Fly.io’s June incident log details five major infrastructure issues: delayed metrics ingestion due to uneven load distribution, elevated CPU and routing failures during a Corrosion database migration,…

Run Paperclip on a Sprite

Fly.io introduced multiple new features for its Sprite product, including running quantized language models (2.4B parameters at 1.58 bits) on Sprites, enabling agent swarms with isolated environments,…

Building Agents that Don't Break Themselves

Fly.io describes a new approach to running AI agents by separating the agent’s durable home (long-lived process) from the execution environment (disposable Sprite sandboxes). Each risky command runs i…

Fly.io forEnterprises

A Public Cloud forSecurityNerds

Meet the TeamsBuilding theFuture

Django Views as Serverless Functions on Fly Machines

Fly.io introduced a method to run Django views as serverless functions using Fly Machines, allowing specific views to offload resource-intensive tasks like PDF generation. The system uses Fly Proxy an…

Staging environments on the Fly with GitHub actions

A guide demonstrates automating staging environments for Django apps on Fly.io using GitHub Actions. Each pull request triggers a temporary staging app with a dedicated Postgres cluster, destroyed aft…

Serverless Functions in Django With Fly Machines

Fly.io introduced a method to run serverless functions in Django using custom commands executed on Fly Machines, leveraging the platform's 'scale to zero' feature. The approach uses Django's built-in …

New goodies in Django 5.0

Django 5.0, now in alpha/beta, introduces three major ORM and form features: generated fields for database-computed values, database-computed default values via Field.db_default, and simplified form f…

Smooth Database Changes in Blue-Green Deployments

Fly.io highlights Django’s SeparateDatabaseAndState() migration operation to enable zero-downtime blue-green deployments. The technique separates model state changes from database schema changes, allo…

Celery Async Tasks on Fly Machines

Fly.io introduced support for running Celery workers on Fly Machines to offload async tasks from Django applications, improving scalability and performance. The guide details setup with Redis as a bro…

Organizing Database Queries: Managers vs. QuerySets

Fly.io’s blog features a technical deep-dive by Mariusz Felisiak on organizing Django database queries using Managers and QuerySets. The post explains how to structure common filters, annotations, and…

QuerySets are Lazy

Fly.io published a technical deep-dive on Django’s lazy QuerySets, explaining how deferred execution reduces unnecessary database queries. The post highlights methods that trigger immediate evaluation…

ASGI deployment options for Django

A Django Fellow details ASGI server options (Daphne, Hypercorn, Uvicorn, Granian) for deploying Django apps, emphasizing performance benefits and Fly.io’s ease of deployment. The post highlights Djang…

How CodeCrafters Helps Devs Upskill Faster

CodeCrafters replaced their in-house Firecracker VM setup with Fly.io’s Machines API to handle variable traffic spikes more efficiently. The switch eliminated over-provisioning costs and reduced engin…

How Lightward Is Betting on Health…and Winning

Lightward, a health-focused creative incubator, attributes its commercial success and team stability to Fly.io’s responsive infrastructure. The company uses Fly.io to support frequent project pivots a…

How MyMahi Moves Fast Enough to Keep up With Teenagers

MyMahi, a New Zealand/Australia-based education platform, expanded its use of Fly.io to handle rapid, unpredictable data syncs from schools. The move replaced AWS Lambda constraints with Fly.io’s glob…

How Tigris Data Makes Object Storage Screaming Fast, Globally

Tigris Data launched a new S3-compatible object storage solution built on FoundationDB, offering real-time global performance by intelligently caching and replicating data near users. It supports smal…