Trending Now - Anthropic Unveils Claude Opus 4.7, Enhancing Software Engineering

WHO:

Mashable - Anthropic is the organization that announced the release of Claude Opus 4.7.

Techmeme - Ashley Capoot is a contributor from CNBC reporting on the announcement.

WHAT:

VentureBeat - Opus 4.7 achieved a 94.2% score on the Graduate-Level Reasoning (GPQA Diamond) benchmark.

VentureBeat - Opus 4.7 scored 91.0% on the Visual Reasoning (arXiv Reasoning) benchmark, up from 84.7% in Opus 4.6.

VentureBeat - The model's success rate in cybersecurity vulnerability reproduction (CyberGym) is 73.1%, compared to Mythos Preview's 83.1% and GPT-5.4's 66.3%.

VentureBeat - The pricing for Opus 4.7 remains at $5 per million input tokens and $25 per million output tokens.

Inc - Claude Opus 4.7 brings sharper coding, stronger reasoning, and a better safety profile.

VentureBeat - Opus 4.7 exceeds OpenAI's GPT-5.4, released in early March 2026, and Google's Gemini 3.1 Pro from February 2026 on key benchmarks.

VentureBeat - Opus 4.7 leads the market on the GDPVal-AA knowledge work evaluation with an Elo score of 1753, surpassing GPT-5.4 (1674) and Gemini 3.1 Pro (1314).

VentureBeat - Opus 4.7 resolved 64.3% of tasks in the Agentic Coding benchmark, compared to 53.4% for Opus 4.6.

VentureBeat - Opus 4.7 achieved a 90.9% score on BigLaw Bench, indicating improved handling of ambiguous document editing tasks.

VentureBeat - The model's visual acuity improved from a 54.5% success rate in visual-acuity tests to 98.5%.

VentureBeat - The updated tokenizer in Opus 4.7 can increase input token counts by 1.0–1.35x.

Mashable - Anthropic released Claude Opus 4.7, a new frontier AI model.

Techmeme - Anthropic stated that Opus 4.7 is "less broadly capable" than Claude Mythos Preview.

Techmeme - Anthropic noted that the "cyber capabilities are not as advanced" in Opus 4.7 compared to Mythos Preview.

Techmeme - Anthropic described Opus 4.7 as a "notable improvement" over Opus 4.6 in advanced software engineering.

Techmeme - Opus 4.7 includes a new "xhigh" effort level.

WHEN:

Mashable - The release of Claude Opus 4.7 occurred on April 16th, 2026.

Techmeme - The announcement regarding Opus 4.7 was made on April 16th, 2026.

WHY:

VentureBeat - Anthropic is launching the Cyber Verification Program to allow cybersecurity professionals to use Opus 4.7 for defensive purposes.

VentureBeat - Anthropic is introducing a new "effort" parameter to allow users to select an xhigh (extra high) effort level for more granular control over the model's reasoning depth.

HOW:

VentureBeat - Opus 4.7 includes a new /ultrareview command designed to simulate a senior human reviewer, flagging subtle design flaws and logic gaps.

Techmeme - Anthropic's Opus 4.7 is positioned as an improvement in software engineering capabilities, indicating advancements in its design and functionality.

(UPDATED April 17, 2026 04:03 UTC)

Opus 4.7 has made headlines by achieving impressive scores, including a 94.2% on the Graduate-Level Reasoning benchmark and a notable increase to 91.0% on the Visual Reasoning benchmark, up from 84.7% in its predecessor. In the realm of cybersecurity, the model shows a success rate of 73.1% in vulnerability reproduction, while Anthropic is launching the Cyber Verification Program to enable professionals to leverage Opus 4.7 for defensive strategies. Additionally, this latest version features a new /ultrareview command that simulates a senior human reviewer, helping to identify subtle design flaws and logic gaps.

(UPDATED April 16, 2026 23:12 UTC)

Anthropic has launched Opus 4.7, which features sharper coding, stronger reasoning, and an enhanced safety profile, surpassing both OpenAI's GPT-5.4 and Google's Gemini 3.1 Pro on key benchmarks. The new model achieved an impressive Elo score of 1753 on the GDPVal-AA knowledge work evaluation and resolved 64.3% of tasks in the Agentic Coding benchmark, marking a significant improvement over its predecessor. Additionally, Opus 4.7 introduces a new "effort" parameter for users to control reasoning depth and includes an /ultrareview command that simulates a senior human reviewer to identify subtle design flaws and logic gaps.

(UPDATED April 16, 2026 21:14 UTC)

Anthropic has announced the release of Claude Opus 4.7, a new frontier AI model, which is described as a notable improvement over its predecessor, Opus 4.6, particularly in advanced software engineering. However, the company also noted that Opus 4.7 is less broadly capable and has less advanced cyber capabilities compared to the Claude Mythos Preview. The announcement and release of this new model took place on Thursday.