Capability change
- Before: Both Deep Research Preview and Deep Research Max Preview were listed with 131,072-token context windows.
- Enabling change: models.dev updated both Apr-21-2026 preview entries to list 1,048,576-token context windows.
- After: The listed limits accommodate research inputs exceeding the previous 131,072-token ceiling, subject to runtime confirmation.
Human consequence: Researchers could keep more source material in one context rather than splitting it into smaller inputs.
Bottleneck: The evidence establishes registry changes, not vendor-confirmed runtime support or effective use of the larger context.
Buildability: B0. Verification: source-inspected. Related shift: shared-context-infrastructure.
Two matching registry updates indicate a context-capacity expansion across both Deep Research previews, rather than merely a price or model-name change.
If runtime support is confirmed, builders could assemble larger research source bundles in a single context, reducing input splitting.
Watch for vendor confirmation and runtime validation: the supplied evidence does not establish availability or research quality at the larger limit.
Published automatically by the New Runtime weekly knowledge desk from the week's evidence.
Supporting evidence
Provenance
Auto-created by the New Runtime weekly knowledge desk on 2026-11-17 from evidence reviewed in the week of 2026-11-09. It is published through the live overlay and joins the hand-edited catalog at the next release.