Anthropic Launches Claude Opus 5 With Enhanced Coding and Autonomous Capabilities

Anthropic released Claude Opus 5 on July 24, 2026, introducing updates for autonomous coding, software development, and scientific research. The model operates at half the cost of Anthropic's flagship Fable 5 model.

Credit: Anthropic

Opus 5 serves as the default model on Claude Max and the top option on Claude Pro. Anthropic kept API pricing equal to the previous Opus 4.8 version. Users can select an effort setting to adjust processing depth, balancing speed against token usage.

Benchmark evaluations put Opus 5 at the top of Frontier-Bench and GDPval-AA for coding and knowledge tasks. The model ranked behind Mythos 5 on cybersecurity tests.

In one Frontier-Bench test, Opus 5 received an image of a machine part without direct visual tools to read the file. The model wrote its own computer vision pipeline, extracted geometry from raw pixels, and generated a 3D FreeCAD model. Competing models failed the task after five attempts.

Enterprise testing showed performance changes across industry workflows. Box reported an 8 percent overall improvement over Opus 4.8. Data analysis tasks increased by 11 percent, and due diligence tasks improved by 17 percent.

Ben Kus, Chief Technology Officer at Box, detailed the testing results:

"Claude Opus 5 delivers the industry intelligence and accuracy that is essential for the analysis of specialized enterprise content. Box found that Opus 5 outperforms Opus 4.8 by 8% and delivers notable performance gains in the data analysis (11% improvement) and due diligence (17% improvement) workflows that technology, healthcare, and public sector organizations rely on daily."

Software testing revealed that Opus 5 inspects its own code outputs. In testing environments, the model rendered user interfaces in simulated desktop and mobile browsers, identified hidden buttons below screen folds, and corrected code before finishing tasks.

Richard Pham, Evals and Product Lead at Anthropic, described the evaluation process:

"Claude Opus 5 checks its own work the way a real frontend developer would. On our benchmark it opened its pages in a browser at desktop and phone widths, caught a product hidden below the mobile fold and an off-screen checkout button, and fixed both before handing the work back."

Engineers reported that the model handles pull requests and code modifications cautious of potential errors.

Credit: Anthropic

Alex Wang, Applied AI engineer, stated:

"Claude Opus 5's judgment is what stands out. Handing off a PR, it doesn't rush to publish: it verifies the branches, checks the template, and thinks through test implications so the handoff is clean. The older models tended to jump ahead and get caught on our checks."

Cristian Rivera, Staff Software Engineer, highlighted changes in codebase updates:

"Claude Opus 5 made large scale changes across our Fundamental Research Assistant codebase, adapting to feedback throughout an agentic workflow and explaining its reasoning more clearly than any model we've used. It handled work we would normally have broken into much smaller pieces."

Fabian Hedin, Co-Founder, described user interface generation:

"Claude Opus 5 is the biggest leap in the Opus family since 4.5. On the same full-stack app builds, the front end shows it first: the best animations, games, and 3D work we have seen from an Opus model."

Niko Grupen, Head of Applied Research, noted performance during multi-step projects:

"Claude Opus 5's biggest gains for us are on longer-horizon work: building a full deck, then revising it. Artifact quality is what decides which model we ship, and this is the clearest step up we've seen — better visual understanding, cleaner formatting, fewer slide issues."

Scientific evaluations showed higher scores in structural biology, bioinformatics, and organic chemistry. Opus 5 scored 10.2 percentage points higher than Opus 4.8 on internal tests inferring molecular structures from spectroscopy data.

Anthropic introduced two beta features alongside the model launch. Developers on the Claude Platform can modify accessible tools mid-conversation without resetting prompt caches.

An automatic fallback feature on the API routes requests flagged by safety classifiers to available models instead of blocking the request.