AI News

Follow the latest AI and automation news

20 Apr 20263 Dhuʻl-Qiʻdah 1447 AH

ByteDance 2025 profit plunges over 70% as AI spending surges

ByteDance reported a significant increase in its AI spending during the third and fourth quarters of last year, resulting in a sharp decline in its net profit margin. The company's profits plummeted by over 70%, highlighting the financial impact of heavy investments in AI. This surge in spending is a strategic move by ByteDance to enhance its AI capabilities, aiming to expand its services and improve user experience. However, these investments come at a critical time as the company faces significant financial pressures due to increasing market competition.

Models

20 Apr 20263 Dhuʻl-Qiʻdah 1447 AH

DeepSeek V4 launch expected this week following researcher's tweet

Social media posts indicate that the highly anticipated launch of the DeepSeek V4 model may be announced this week, following a tweet from a researcher in the field, generating excitement within the AI community. DeepSeek V4 is one of the most awaited large models in the industry, expected to deliver significant performance improvements over previous versions. Its release could transform the competitive landscape in the large model market, providing new opportunities for developers and companies while enhancing AI applications and fostering innovation.

Business

18 Apr 20261 Dhuʻl-Qiʻdah 1447 AH

DeepSeek reportedly launches first funding round at over $10 billion valuation

DeepSeek plans to raise $300 million in external funding after previously rejecting venture capital. This funding round comes with a valuation exceeding $10 billion, showcasing market confidence in its technological capabilities. The company is known for its advanced AI solutions, aiming to expand its services and enhance its growth. This move is significant as it positions DeepSeek to compete more effectively in the AI sector. The funding could lead to new opportunities for innovation and development within the company, potentially impacting the broader AI landscape positively.

💡

Alibaba open-sources Qwen3.6-35B-A3B to rival larger AI models in coding

What implications does the Qwen3.6-35B-A3B model have for AI development?

Models

Open-world evaluations for measuring frontier AI capabilities

16 Apr 202628 Shawwal 1447 AH

Models

16 Apr 202628 Shawwal 1447 AH

Open-world evaluations for measuring frontier AI capabilities

Researchers have begun testing artificial intelligence in real-world environments, introducing the term "open-world evaluations." This approach aims to measure models' abilities to create real products or conduct scientific experiments. In the first trial, an AI agent developed an iOS application with only two errors, indicating useful potential but also possible risks. Seventeen researchers from various fields collaborated on the CRUX project to assess AI capabilities through these evaluations.

Agents

Huawei unveils OfficeClaw to capture China's growing desktop AI market

16 Apr 202628 Shawwal 1447 AH

Agents

16 Apr 202628 Shawwal 1447 AH

Huawei unveils OfficeClaw to capture China's growing desktop AI market

Huawei has launched OfficeClaw, an AI system designed to enhance office scenarios. OfficeClaw facilitates content creation, intelligently organizes local files, and quickly prepares presentations. This launch coincides with the rising demand for desktop AI solutions in the Chinese market. With OfficeClaw, Huawei aims to address the needs of users in modern work environments, enhancing productivity through advanced tools. The system features innovative capabilities such as automatic content generation, allowing users to concentrate on core tasks.

Models

Alibaba launches Happy Oyster world model for real-time 3D creation

16 Apr 202628 Shawwal 1447 AH

Models

16 Apr 202628 Shawwal 1447 AH

Alibaba launches Happy Oyster world model for real-time 3D creation

Alibaba has launched the Happy Oyster model, enabling the creation of dynamic three-dimensional environments. This model offers technical support for various scenarios, including film production and game development. Happy Oyster's capability for real-time 3D creation allows developers and artists to work more efficiently. This technology marks a significant advancement in 3D modeling, facilitating greater interaction with content.

Agents

Chinese embodied AI startup Tars raises $455 million in Pre-A funding round

16 Apr 202628 Shawwal 1447 AH

Agents

16 Apr 202628 Shawwal 1447 AH

Chinese embodied AI startup Tars raises $455 million in Pre-A funding round

Chinese startup Tars has raised $455 million in a Pre-A funding round, setting a record for the highest single financing in the embodied AI sector. This funding round attracted notable investors, showcasing strong market confidence in embodied AI technologies. Tars plans to utilize the funds to expand its operations and develop new technologies in embodied AI, which encompasses robotics and autonomous systems. This investment comes at a time of increasing global interest in AI and its applications in everyday life.

Models

My bets on open models, mid-2026

15 Apr 202627 Shawwal 1447 AH

Models

15 Apr 202627 Shawwal 1447 AH

My bets on open models, mid-2026

We are at a pivotal moment to see if open models can keep pace with closed labs. The clear answer is no, as the capability gap between models is becoming increasingly complex. This dynamic is intertwined with funding for open models, who builds them, and techniques that enable fast-following. The capability gap is a key signal amid complex forces shaping supply and demand. Demand for open models is high, with many individuals, organizations, and sovereigns seeking them, while supply is dictated by economics. The question of which business strategies support releasing open models remains open.

Models

Gemini 3.1 Flash TTS: the next generation of expressive AI speech

15 Apr 202627 Shawwal 1447 AH

Models

15 Apr 202627 Shawwal 1447 AH

Gemini 3.1 Flash TTS: the next generation of expressive AI speech

The company has launched its latest audio model featuring granular audio tags. These tags provide users with enhanced control over AI-generated speech, allowing for more expressive audio outputs. This model marks a significant advancement in audio processing and user experience. The tags improve the accuracy of speech generation by allowing greater control over tone and rhythm. This development is crucial amid increasing competition in the AI audio space and opens new applications in gaming and entertainment.