Anthropic's Claude Opus 4.7 Revolutionizes Coding Capabilities with 64.3% SWE-Bench Pro Score

The latest iteration of Anthropic's Claude model, Opus 4.7, has made a substantial impact in the world of autonomous coding, achieving an impressive 64.3% score on the SWE-bench Pro coding benchmark. This represents a significant improvement over its predecessor, Opus 4.6, which scored 53.4%, and even surpasses OpenAI's GPT-5.4, which scored 57.7%. Although Anthropic's own top model, Claude Mythos Preview, still leads with a wide margin at 77.8%, Opus 4.7's advancements are a major step forward for the company.

One of the key features of Opus 4.7 is its ability to follow instructions more precisely than its predecessor, which may lead to unexpected results for prompts written for older models. This increased precision is a double-edged sword, as it allows for more accurate coding but may require developers to rework their existing prompts. Additionally, Opus 4.7 boasts triple the image resolution of earlier Claude models, processing images at up to 2,576 pixels on the long edge, equivalent to roughly 3.75 megapixels. This enhancement is particularly beneficial for computer-use agents that need to read dense screenshots and extract data from complex diagrams.

The improved image resolution has also led to significant gains in various benchmarks, including the Document Reasoning benchmark (OfficeQA Pro), where Opus 4.7 achieves 80.6% accuracy, up from 57.1% with Opus 4.6. Furthermore, the model demonstrates notable advancements in biomolecular reasoning and visual navigation (ScreenSpot-Pro). These enhancements make Opus 4.7 an attractive option for developers and businesses looking to leverage AI for complex tasks.

In a deliberate effort to mitigate potential risks, Anthropic has throttled the model's cybersecurity capabilities, experimentally reducing certain cyber capabilities during training. This move is likely a response to growing concerns about the potential misuse of AI models for malicious activities. While this may limit the model's capabilities in certain areas, it also ensures a safer and more responsible AI ecosystem.

The pricing for Opus 4.7 remains the same as its predecessor, with per-token prices unchanged. However, a new tokenizer maps the same text to up to 35% more tokens, which can significantly increase the actual cost per request. This may be a consideration for developers and businesses looking to integrate Opus 4.7 into their workflows.

Historically, Anthropic's Claude models have been known for their impressive coding capabilities, and Opus 4.7 is no exception. The model's ability to learn from feedback and adapt to new tasks has made it a popular choice among developers. With Opus 4.7, Anthropic has taken a major step forward in autonomous coding, and its implications are far-reaching. For developers, this means more accurate and efficient coding, while businesses can leverage the model's capabilities to streamline their operations and improve productivity.