Claude 4 Era Ends: Early Notes From the Migration

The Claude 4 generation is now officially retired. One day past the June 15 deadline, early migration data suggests most teams moved in time — and reveals which preparation habits actually paid off.

Claude 4 Era Ends: Early Notes From the Migration

June 15 has passed, and Claude Sonnet 4 and Opus 4 are now retired. Requests pinned to the old identifiers began failing at the deadline, making this week a natural checkpoint for how the industry handled one of the largest forced model migrations to date.

The early picture: teams that treated the retirement as a scheduled project moved smoothly. Most had inventoried their integrations in May, run evaluation suites against Sonnet 5 and Opus 4.7 in parallel, and flipped traffic over gradually behind routing layers. For these organizations, June 15 was a non-event — the best possible outcome for infrastructure work.

The stragglers shared recognizable patterns. Some discovered undocumented dependencies on 4-series output formatting that their downstream parsing assumed. Others had evaluation suites that existed on paper but had never run against production traffic, turning revalidation into archaeology. A few simply missed the announcement — a reminder that model deprecations now arrive through vendor changelogs the way security patches do, and deserve the same monitoring.

The capability news is mostly pleasant. Migrating teams generally report the 5-series and late 4-series models outperform what they replaced on coding and multi-step tasks — Opus 4.7's 64.3% on SWE-bench Pro earlier this year set the tone — though capability jumps bring their own surprises in refusal patterns and formatting habits.

Our recommendation going forward: maintain a living model-migration runbook. With at least one more 5-series release expected this summer and rivals on comparable cadences, the next retirement is already visible on the horizon. The teams that documented this week's process just made their next migration cheaper.