Claude Opus 4.7: What Changed, Pricing, and API Name

Claude Opus 4.7 is Anthropic's most capable generally available model as of April 16, 2026, and the official API model name is claude-opus-4-7. Anthropic's Claude Opus 4.7 launch post and Anthropic's current models overview both confirm that pricing stays at $5 per million input tokens and $25 per million output tokens while the release focuses on stronger coding, vision, and long-running multi-step work.

What Anthropic launched

Anthropic launched Claude Opus 4.7 on April 16, 2026 and described it as generally available on the same day in Anthropic's Claude Opus 4.7 launch post. The company says the model is stronger than Opus 4.6 on advanced software engineering, difficult multi-step tasks, higher-resolution vision, and professional work like interfaces, slides, and documents.

The product page at the Claude Opus product page uses almost the same framing but adds the practical positioning: Opus 4.7 is for professional software engineering, complex agentic workflows, and high-stakes enterprise work where prior models still needed too much supervision.

That matters because this is not a cheap-model story. Anthropic is not trying to turn Opus 4.7 into the default model for every request. It is trying to make the expensive flagship more useful for the hardest tasks that teams already reserve for Opus-tier models.

Pricing, IDs, and core specs

Anthropic's current documentation now lists the concrete details that most teams need before they can even benchmark the release. The fastest way to verify the live launch facts is still Anthropic's current models overview, because that page shows the current model IDs and limits rather than only the marketing copy.

What changed from Opus 4.6

The cleanest official summary is simple: Opus 4.7 keeps Opus-level pricing but is supposed to be more useful on the work people actually escalate to Opus. Anthropic's Claude Opus 4.7 launch post says the biggest launch-day gains are in advanced software engineering, instruction-following, self-verification, and higher-resolution vision.

The updated Anthropic's current models overview also exposes one subtle but important product change. It lists adaptive thinking for Opus 4.7, but not the older "extended thinking" label. That suggests Anthropic wants developers to treat 4.7 as a newer reasoning model generation rather than just Opus 4.6 with a marginal benchmark bump.

For direct historical context, the archived Claude Opus 4.6 product page shows how Anthropic positioned Opus 4.6 at launch: same $5/$25 token pricing, 1M context on the product page, and a flagship pitch around coding, agents, and enterprise workflows. The release message for 4.7 is not that the price got better. The message is that the same premium slot now buys a stronger coding-and-agents model.

Where Claude Opus 4.7 is available

Anthropic says Opus 4.7 is available across Claude, the Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. the Claude Opus product page and Anthropic's Claude Opus 4.7 launch post both say that directly, and the current Anthropic's current models overview repeats the platform list while also publishing the live API ID.

There is one nuance worth calling out. The same Anthropic's current models overview page adds a footnote that Claude Opus 4.7 on AWS is currently available through Claude in Amazon Bedrock in research preview. So the launch is real, but Bedrock rollout details still matter if you need immediate production access through AWS rather than Anthropic's own API.

If you just want the safest launch-day move, use the first-party Anthropic API first. If you need cloud procurement, regional controls, or consolidated vendor billing, then verify the exact platform status in Bedrock, Vertex AI, or Foundry before migrating workloads.

What to watch next

The two things to watch next are the safety documentation and the independent benchmark picture. Anthropic's Anthropic's system cards page still listed Opus 4.6 and older system cards when I checked after launch, which means the release pages were live before the public system-card index had fully caught up.

The other thing to watch is whether public benchmark trackers, IDE vendors, and agent platforms reproduce Anthropic's launch-day claims in broader testing. Right now the official story is strong enough to justify immediate evaluation, but not strong enough to skip your own benchmark harness.

Limitations and Tradeoffs

The launch facts are clear, but some of the strongest performance claims are still Anthropic-led or customer-led evals rather than neutral third-party benchmarks. If you already have Sonnet 4.6 or Opus 4.6 working in production, do not swap models blind just because the launch copy sounds strong. Benchmark your own coding, agent, and document workflows first.

Related Guides

• Best Claude Models in 2026
• Claude Opus 4.6 on OpenClaw
• Claude Mythos and Project Glasswing
• OpenClaw vs Claude Pro

FAQ

What is the Claude Opus 4.7 API name?

The live Anthropic models overview lists the API identifier as `claude-opus-4-7`. That is the model name you should use in first-party Anthropic API requests.

How much does Claude Opus 4.7 cost?

Anthropic lists Claude Opus 4.7 at $5 per million input tokens and $25 per million output tokens. The company says pricing remains the same as Opus 4.6.

What is the Claude Opus 4.7 context window?

Anthropic's current models overview lists a 1M-token context window for Claude Opus 4.7, plus 128k max output on the synchronous Messages API.

Is Claude Opus 4.7 available outside Anthropic's own API?

Yes. Anthropic says Claude Opus 4.7 is available on Claude itself, the Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry, although the docs note that the AWS route is currently research preview.