Microsoft’s new Xbox chief has had a busy couple of months after promising “the return of Xbox.” Asha Sharma met with publishers at the Game Developers Conference in March, and has also been on the road visiting Microsoft’s own game studios and product teams in recent weeks. Sharma, who used to work in Microsoft’s CoreAI […]
Microsoft Plans Surface Laptop with OLED Display
Microsoft is preparing to launch new Surface Pro and Surface Laptop models in the coming months. Windows Central reports that the refreshed models will include both Intel and Qualcomm variants, with Microsoft opting for an OLED display option for the Surface Laptop this year. Microsoft is expected to launch new Surface Pro and Laptop models […]
Google Integrates Nano Banana-Powered Image Generation into Gemini’s Personal Intelligence
The feature will be initially available to paid users in the U.S.
Roblox’s AI Assistant Gains New Agentic Tools to Plan, Build, and Test Games
The new new tools are designed to help creators throughout the entire development process.
AI Traffic to US Retailers Increased by 393% in Q1, Boosting Revenue
Adobe says AI traffic to U.S. retail sites also jumped 269% in March, with visitors converting better and generating more revenue than non-AI shoppers.
InsightFinder Secures $15M to Aid Companies in Identifying AI Agent Errors
According to CEO Helen Gu, the biggest problem facing the industry today is not just monitoring and diagnosing where AI models go wrong, it’s diagnosing how the entire tech stack operates now that AI is a part of it.
Two Americans Sentenced for Assisting North Korea in $5 Million Fake IT Worker Scheme
The U.S. Department of Justice announced that two Americans were sentenced to years in prison for helping the North Korean government place fake IT workers in U.S. companies.
Google Pixel 10 Gets Major Security Enhancement Advised for Every Android Device
Google’s Pixel 10 series devices are receiving a significant enhancement in security, as the company has recently initiated a substantial security upgrade that will soon reach your device. In a post on Google’s Security Blog from April 2026, the Android manufacturer announced that it has fortified its proactive security strategies on the Pixel 10 by “embedding a memory-safe Rust DNS (Domain Name System) parser into the modem firmware.”
This provides Pixel 10 series devices with an essential security boost, as Google indicates that utilizing the Rust programming language in its baseband DNS parser mitigates memory-related vulnerabilities, which consequently helps to greatly minimize security threats in the modem. Given the severe nature of memory-related vulnerabilities, one can hope that other Android phone makers will adopt Google’s approach.
Reasons for other Android manufacturers to follow suit
Claude Opus 4.7 Outperforms GPT-5.4 and Gemini 3.1 Pro on SWE-bench and Agentic Reasoning
In short: Anthropic has released Claude Opus 4.7, its most capable generally available model, with benchmark-leading scores on SWE-bench Pro (64.3% vs GPT-5.4âs 57.7%), multi-agent coordination for hours-long workflows, 3x higher image resolution, and a 14% improvement in multi-step agentic reasoning with a third of the tool errors. Priced at $5/$25 per million tokens, it […]
This story continues at The Next Web
Anthropic Launches Opus 4.7 Model Highlighting Sophisticated Software Engineering Methods
**Anthropic Launches Claude Opus 4.7: Progress in AI for Software Development**
Anthropic has revealed its newest AI model, Claude Opus 4.7, only two months following the release of Opus 4.6. This updated version focuses on enhanced software development features, representing a notable advancement in managing intricate coding activities.
### Opus 4.7: Superior Performance in Software Engineering
Claude Opus 4.7 is tailored to require less oversight for demanding coding activities. Users have indicated that they can confidently assign their most challenging coding tasks to Opus 4.7, which shows improved thoroughness and reliability in handling complex, lengthy operations. The model is attentive to directives and has developed methods to validate its outputs prior to delivery.
### Boosted Vision and Creativity
Anthropic asserts that Opus 4.7 incorporates significantly improved vision abilities, enabling it to analyze images at elevated resolutions. Furthermore, the model displays greater creativity and discernment when executing professional assignments, yielding higher-quality results such as user interfaces, presentation slides, and documentation.
### Benchmark Performance
In head-to-head evaluations, Opus 4.7 surpassed its forerunner, Opus 4.6, as well as other models like GPT-5.4 and Gemini 3.1 Pro. However, it still lags behind the more adaptable Claude Mythos Preview, which is not yet broadly accessible to the public and is currently only being shared with a limited number of software platform vendors.
### Key Enhancements
Anthropic emphasizes several significant upgrades in Opus 4.7, such as:
– **Instruction Following**: Enhanced capacity to comply with user directives.
– **Multimodal Support**: Boosted abilities in handling diverse types of data.
– **Memory Utilization**: Improved management of file system-based memory, enabling the model to retain vital notes across various sessions, thereby minimizing the necessity for extensive context in new tasks.
### Consistent Upgrade Schedule
Anthropic has implemented a regular upgrade timetable for its Claude Opus model, delivering updates every two months. Opus 4.7 closely follows Opus 4.6, which itself was launched two months post Opus 4.5.
### Token Usage Modifications
With the launch of Opus 4.7, there are significant alterations in token usage. The model uses a new tokenizer that enhances text processing but may lead to increased token counts for identical inputs. Additionally, Opus 4.7 functions at higher effort levels, especially in complex situations, resulting in a greater number of output tokens.
### Claude Code Enhancements
In conjunction with the new model, Anthropic has made incremental improvements to Claude Code, part of the Claude Mac application. Importantly, the auto mode is now accessible to Max plan subscribers, broadening availability beyond Teams/Enterprise/API customers. Moreover, a new command, “/ultrareview,” has been launched to support focused review sessions that pinpoint potential issues in code modifications.
For additional information, users can check Anthropic’s blog post and the Claude Opus 4.7 System Card available on their website.
