Mavericks AI News Passes 80,000 Subscribers and Publishes Its December 22 Issue

Mavericks, Inc., the developer of the video generation AI NoLang and other advanced AI products, has published the latest issue of Mavericks AI News, a generative AI newsletter that has passed 80,000 subscribers and is one of Japan’s largest¹. The issue was distributed in the week of December 22. It reviews the practical value of Gemini 3 Flash, the very fast, low-cost model announced by Google, by comparing benchmark results; the audio separation technology SAM Audio released by Meta; and the latest moves by ChatGPT, which has strengthened integrations with external applications such as Photoshop and is turning itself into an all-purpose platform. Drawing on the knowledge of an engineering team that works at the front line of AI development every day, the newsletter goes beyond feature summaries and digs into how each development can actually be used at work.

One of Japan’s largest, with 80,000 subscribers: Mavericks AI News, written by a generative AI product team

Mavericks, Inc. is a Japanese startup that offers the generative AI product NoLang. Within one year of launch, NoLang passed 150,000 registered users and has been adopted by more than 60 corporate customers, and the team operates its own product daily at the front line of AI-driven development. The generative AI industry changes extremely fast, so the questions keep coming: what can generative AI do right now, which AI services are drawing attention in the market, and what kind of AI will arrive next. Mavericks AI News, distributed every Monday, collects the topics picked by a team that follows product development at close range.

In Mavericks AI News, Mavericks selects information that is substantive and directly useful at work, seen from the perspective of a team that builds AI products every day. The newsletter does not stop at introducing technology. It aims to help readers, including those with no prior knowledge of AI, picture concretely how they could put each development to use tomorrow. Each story pairs the latest generative AI news with original commentary grounded in hands-on expertise in AI product development.

Highlights from this week’s issue, distributed on December 22

Here is a short introduction to the topics covered in the December 22 issue.

Feature 1: Google releases its new Gemini 3 Flash model, but what problems does Gemini still have?

On December 17 US time, Google released the compact AI model Gemini 3 Flash. It is designed around speed and cost performance. In benchmarks it matches Gemini 3 Pro and in some cases beats it, while running three times faster than Gemini 2.5 Pro at roughly one tenth of the effective cost. According to Google, it is already built into services across the company, including the Gemini app and AI Mode in Google Search.

The issue covers the specifications of this fast, cost-efficient new model in detail, and gives a frank review, with concrete examples, of the practical weaknesses that Gemini still shows against competitors in search and coding.

Feature 2: Meta releases SAM Audio, which isolates only the sound you want from audio and video

Meta has released SAM Audio, an AI model that extracts only a specific sound from audio or video. It can separate a piece of music into individual instruments, or pull a clear speaking voice out of a noisy interview video. Alongside this striking capability, which lets users separate audio intuitively through natural language or a click, the issue explains the broader trend of AI models advancing across modalities: images, video, 3D and audio.

Feature 3: ChatGPT updates its apps feature, so it can now call Photoshop and Acrobat

ChatGPT recently updated its apps feature, and the number of applications that can be used directly inside ChatGPT grew significantly. Through the integration with Adobe products, for example, image editing in Photoshop and PDF editing in Acrobat can now be completed from ChatGPT without switching screens. OpenAI is working to turn ChatGPT from a chat AI into a platform where every task can be finished in one place. The issue examines where the ChatGPT App Store concept stands, accelerated by the Adobe integrations, the link with the app-building tool Lovable and the release of the Apps SDK, and considers the potential and the open questions of that shift to a platform.

Looking ahead

As a product-out generative AI startup, Mavericks, Inc. sees returning to society the knowledge it gains from developing advanced products in-house as an important part of its mission. Through Mavericks AI News, the company will continue to deliver reliable, practical information to every business professional, developer and digital transformation lead working through the AI era.