Lifecycle intelligence for AI stacks

Model shutdowns should fail in CI, not production.

ModelShift finds deprecated OpenAI, Anthropic, Gemini and Bedrock model IDs in your code — then shows the deadline and migration target.

Try the live scanner
  • Runs locally
  • Zero source uploads
  • No runtime dependencies
modelshift / production-api

$ modelshift . --fail-on high

ModelShift 0.2.0 — 2 lifecycle risk(s)

CRITICAL
claude-sonnet-4-20250514retired · migrate to claude-sonnet-4-6
src/agent.py:18
HIGH
gpt-4-0613EOL Oct 23 · migrate to gpt-5.6-sol
src/router.ts:42

Process exited with code 2 · deployment blocked

40tracked model IDs
7providers
3CI-ready formats
0files uploaded

Live product demo

Paste code. See the risk instantly.

The demo runs entirely in this page. Your snippet never leaves your browser.

Code input
Local-only analysis
FindingsReady

Ready to inspect

Run the sample to identify lifecycle risks and replacements.

Built for the path to production

One scanner. Three control points.

01

Developer laptop

Scan a repository locally without sending source code or secrets to a hosted service.

modelshift .
02

Pull request

Return a non-zero exit code when critical or high-risk model references enter the diff.

--fail-on high
03

Security tooling

Export normalized findings for automation, dashboards and code-scanning workflows.

--format sarif

Source-backed lifecycle registry

Deadlines change. Your code usually doesn’t.

Each tracked ID maps to the provider’s official lifecycle page, an exact date when published, and a recommended replacement when one exists.

Manually verifiedSeptember 18, 2026
Next deadlinesLive registry

Loading verified lifecycle data…

Simple path from scan to control

Start free. Automate when the signal proves useful.

Paid plans are proposed for validation and are not yet available for purchase.

CLI Preview

€0 forever
  • Local repository scan
  • Text, JSON and SARIF
  • CI failure policies
  • Verified starter registry
Try in browser

Team Policy

€99 / org / month
  • Multi-repository inventory
  • Central risk policies
  • Ownership and audit trail
  • Team integrations

Straight answers

Before you add it to CI.

Does ModelShift upload my code?+

No. The CLI reads files on the machine where it runs and reports only matched model identifiers, paths and line numbers. The browser demo also runs locally in your browser.

Can it migrate models automatically?+

Not in the preview. ModelShift identifies lifecycle risk and known replacements. Automated migration needs provider-specific evaluations and human approval before it can be trusted.

How accurate is the registry?+

Entries are sourced from official provider lifecycle documentation. Providers can revise dates, so every release records a verification date and links back to the source.

Is this already for sale?+

No. This is a private pre-launch build. Pricing is a validation hypothesis; no payment is accepted before the operator’s legal structure is active.

Private pre-launch build

Know which model breaks next.

Run the demo now. Enable public access and paid pilots only after the company is legally ready.