ToolsThe story, in brief

Python function bundles now include precompiled bytecode

2.8s to 1.3s. Vercel just cut Python cold starts in half with precompiled bytecode.

Paper-cut illustration of a coral software window opening into a three-dimensional drafting space.
New tools for building and creating with AI.AI illustration by KeyNews
The KeyNews take

Why it matters

Vercel's bytecode precompilation is a meaningful infrastructure optimization for AI/ML workloads running on serverless. For teams deploying inference functions or data pipelines, 53% cold-start reduction directly impacts latency-sensitive applications and operational costs.

The key facts

11 to know
  1. Cold start time reduced from 2.8s to 1.3s (53% improvement)

  2. Precompilation happens at build time, not runtime

  3. Applies to both application code and dependencies

  4. .pyc files bundled automatically

  5. No code changes required from developers

  6. Larger functions near bundle size limit see smaller gains

  7. Cold start reduction: 2.8s → 1.3s (53% improvement for median-sized functions)

  8. Feature: Automatic .pyc bytecode compilation at build time for application code and dependencies

  9. No code changes required for developers

  10. Functions near size limits see smaller gains due to bundle constraints

  11. Published July 20, 2026

Go to the source

Vercel Blogvercel.com

Publisher excerpt: Vercel now compiles Python functions to bytecode at build time. In our benchmarks, cold starts for the median-sized function dropped from .2.8s to 1.3s When Python imports a module without cached bytecode, it parses and compiles the source before executing it. That compilation step adds startup…
Read original report
Back to today's editionMore tools news

The wider picture

View all
Paper-cut illustration of a coral software window opening into a three-dimensional drafting space.
AI illustration by KeyNews
Tools01

OpenAI nabs key Patreon execs ahead of upcoming announcement

OpenAI is building a creator-focused product suite with deep domain expertise (Patreon's co-founder + product + engineering leads). This signals a major new revenue and engagement vector for ChatGPT — and a direct threat to Patreon's existing creator economy.

The Verge AI
Illustration of a transparent lens revealing connected networks across layers of paper.
AI illustration by KeyNews
Tools02

Nokia Open-Sources AnyJev: A Training-Free Layer That Turns Any Open LLM Into a Calibrated Decision Model

A practical, deployment-ready layer for a common production pattern — classification over generation — that practitioners can drop into existing LLM stacks immediately. No fine-tuning required.

MarkTechPost
Paper-cut illustration of a coral software window opening into a three-dimensional drafting space.
AI illustration by KeyNews
Tools03

SpeakON Ships a MagSafe AI Voice Button With Its Own Microphone

A hardware-first approach to voice AI workflow — MagSafe button with onboard mic addresses the real friction in voice-to-text-to-action. Relevant to practitioners building voice UX and to the broader consumer AI tooling wave.

MarkTechPost