Update serverless Python hosting comparison with 2024-2026 platform changes

Merged
HB
4mo ago
+193-48

Summary

Comprehensive update to the serverless Python hosting comparison document to reflect significant platform changes, new entrants, and pricing/feature updates through July 2026.

Key Changes

  • Added "Recent Platform Changes (2024–2026)" section documenting major updates across all platforms:

    • AWS Lambda: Python 3.14 GA, cold-start billing, new credit-based free tier
    • Azure Functions: Flex Consumption plan GA with 1,000 instance scaling
    • Google Cloud: Cloud Run functions rebranding, Python 3.14 with uv default
    • Cloudflare Workers: Python Workers redux with 10x faster cold starts via memory snapshots
    • Vercel: Fluid Compute with Active CPU pricing, Python 3.13/3.14 support
    • Alibaba Cloud: Switched to tiered Compute-Unit billing, free tier replaced with trial
    • Tencent Cloud: Free tier eliminated, API Gateway triggers decommissioned
    • Fly.io: Free allowances removed for new orgs, GPU deprecation notice
    • Fermyon: CNCF Sandbox membership, Akamai acquisition
    • Oracle OCI: Long-running functions up to 1 hour, 3GB max memory
    • IBM Code Engine: Python 3.13 support, 3.11 deprecation
    • Render: New flat-fee workspace plans with reduced included bandwidth

  • Added four new platform entries:

    • Modal (Python-native serverless with decorator-based API)
    • DigitalOcean Functions
    • Scaleway Serverless Functions
    • Koyeb (container-based scale-to-zero)

  • Updated DevEx table with current Python versions, status, and framework support across all platforms

  • Revised Pricing table with:

    • Current free tier details (many platforms eliminated ongoing free tiers)
    • Updated first paid tier pricing
    • Clarification of billing models (per-request vs per-second instance billing)
    • Added notes on trial-only free tiers and credit-based plans

  • Expanded Runtime Limits table with more detailed specifications:

    • Added default/max memory configurations
    • Clarified execution time limits and payload sizes
    • Added scale limits and concurrency information
    • Included notes on CPU allocation and billing models

  • Updated Other Platform Products table with new services and corrections:

    • Added database, cache, storage, and AI/ML offerings for new platforms
    • Updated existing platform capabilities (e.g., Fly.io Tigris, Vercel Blob)

  • Enhanced Performance section with cold-start data for new platforms

  • Expanded Discussions/Community/Support table with entries for new platforms

  • Added comprehensive References section with official documentation links for all platforms, organized by provider

  • Updated metadata: Data verification date (July 2, 2026), added disclaimer about pricing subject to change, clarified distinction between true FaaS and scale-to-zero container platforms

Notable Implementation Details

  • Maintained consistent table formatting and emoji indicators (✅/🚫/❔/👍/🎉) across all sections
  • Added contextual notes explaining platform transitions (e.g., Tencent maintenance mode, Alibaba CU billing)
  • Included links to official documentation and release notes for all major changes
  • Clarified billing model differences (wall-clock vs CPU time, per-request vs per-second)
  • Added GPU-first platform callout (RunPod, Beam, Cerebrium) as out of scope but relevant

https://claude.ai/code/session_019v36xvky971cALdxua1GZ9

Summary by CodeRabbit

  • Documentation
    • Refreshed the Python serverless host comparison guide with updated verification date and broader provider coverage.
    • Revised comparison tables for developer experience, pricing, runtime limits, performance, support, and platform capabilities.
    • Updated excluded-platform messaging and links so the document reflects current availability and references.