Major Players

NTT Docomo Business held a strategy announcement to explain its 2030 revenue targets and growth strategies for strengthening AI implementation capabilities.

• •A business strategy announcement titled "Taking Corporate AI Implementation to the Next Growth" was held on September 28, 2026.
• •Three growth strategies were presented: an AI-optimized platform, four priority areas, and partnering.
• •A roadmap toward a 2.6 trillion yen corporate revenue target for fiscal 2030 was disclosed.

Google announced its new AI model, Gemini 4 Argon, claiming it outperforms competitors in many areas according to internal evaluations.

• •Google announced its new AI model, Gemini 4 Argon.
• •Internal evaluations claim it outperforms Astra, Fable 5.1, and Opus 5.5 in many areas.
• •It will initially be provided to cyber defense organizations, with general availability planned as soon as possible.

GUGA announced a major revision to the Generative AI Passport syllabus, effective from the February 2027 exam.

• •GUGA significantly revised the syllabus for the Generative AI Passport qualification exam, which helps prevent generative AI risks.
• •The revised syllabus will apply to exams held from February 2027.
• •Pre-orders for the revised official textbook and acceptance for qualification renewal tests have begun.

Cloudflare released two open-source multimodal decision-making models trained in-house based on Qwen.

• •Cloudflare released two decision-making models, "Clef" and "Clef-flash," on October 1, 2026.
• •The models were trained internally by Cloudflare using Qwen as the base.
• •They are provided as open source on Hugging Face under the Apache 2.0 license.

Microsoft AI released low-latency streaming transcription and tone-preserving multilingual speech synthesis models, starting previews on Microsoft Foundry for voice agents.

• •MAI-Transcribe-2-Streaming, a streaming transcription model supporting low-latency sequential recognition, was announced.
• •A suite of speech synthesis models supporting multilingual synthesis while preserving voice tone was simultaneously released.
• •Preview availability began on Microsoft Foundry, accompanied by composite demos designed for building voice agents.

Amazon's in-car device 'Echo Auto' has disappeared from official search, suggesting a possible end of sales.

• •Amazon's in-car device 'Echo Auto' no longer appears in search results on the company's website.
• •The inability to search for the product has led to speculation that sales may be ending.
• •Specific end-of-sale dates or information about replacement products are not provided in the article.

NTT TechnoCross launched a new version of its AI visual regression testing tool AIspector, achieving screen comparison close to human visual inspection.

• •The new version of AIspector, a tool that automates screen differences and operation checks with AI, became available on October 1.
• •It features a smart approximate comparison function that automatically tolerates minor position shifts or color changes that are not practical issues.
• •By supporting efficient screen comparison close to human visual perception, it improves test accuracy and efficiency.

SIE and AMD announced QSSR, an AI upscaling technology for the standard PS5 that improves image quality.

• •SIE announced QSSR, a lightweight version of PSSR developed with AMD to enable AI-based image quality improvements on the standard PS5.
• •The technology will be introduced first in titles like Marvel's Wolverine, with availability expanding to all developers in the future.
• •Ghost of Yōtei is also included among the initial supported titles.

Apple's next smart home camera reportedly adopts a different approach than Ring or Google Nest, designed to understand home events without recording video.

• •Apple's next smart home camera is designed to understand what is happening in the home.
• •Reports indicate it adopts a very different approach from Ring or Google Nest products.
• •It is suggested that the device does not record video to protect privacy.

OpenAI added virtual try-on and document scanning features to ChatGPT to enhance the user experience.

• •A virtual try-on feature was added, allowing users to generate images of themselves wearing clothes or accessories using their photos from product information cards.
• •A favorites feature was also included to organize items of interest.
• •The iOS app introduced a scanning feature that automatically compiles consecutively photographed documents into a single PDF for sending.