Anthropic shipped a cluster of changes around Claude Code and its API that collectively look less like a single feature launch and more like workflow sanding: fewer special cases, fewer knobs, and better default behavior for long-context coding and multimodal requests.
In a Friday update, Anthropic’s Alex Albert listed four concrete changes: Opus 4.6 1M is now the default Opus model for Claude Code users on Max, Team, and Enterprise plans; the API no longer has a long-context price increase; the API no longer requires a beta header; and requests can now include up to 600 images.
Opus 4.6 1M becomes the default in Claude Code (for some plans)
The most visible end-user shift is that Opus 4.6 1M is now the default Opus model in Claude Code for Max, Team, and Enterprise subscribers. In practical terms, that signals Anthropic is comfortable making its long-context option the “normal” path in its own coding tool—no manual model hunting required.
Developers in the replies framed it as an immediate quality-of-life upgrade for long-running sessions and large projects, with less need to compress context mid-stream.
The API drops the long-context pricing premium
The change likely to matter most for production usage is pricing: “No more long context price increase in the API.” Several responses homed in on economics rather than raw capability—arguing that charging a premium at the long end effectively discouraged the very overflow-the-window use cases that long context was meant to support.
One reply described prior long-context pricing as a pressure to keep prompts artificially short, and another noted teams had been “rationing” large-context usage on bigger repos.
No beta header required
The other API-focused cleanup is operational: “No beta header required in the API.” That’s small on paper, but it removes a recurring source of friction for automated pipelines and stricter deployment processes—one less bespoke requirement to carry through SDK wrappers, gateways, and policy reviews.
Up to 600 images per request
Finally, Anthropic is raising the ceiling on multimodal input: up to 600 images in a single request. Some replies immediately pointed to document-heavy workflows—large batches of screenshots, mixed docs, and visual references—where splitting inputs across calls can become its own kind of prompt engineering.
As always, the practical limit may be less about the API and more about the surrounding system’s ability to assemble and ship that payload reliably.
Source: https://x.com/alexalbert__/status/2032522722551689363
