AI News

Follow the latest AI and automation news

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Thirst for local AI chips raises revenue for China GPU leaders Cambricon, MetaX

Cambricon and MetaX reported significant revenue increases in Q1, with Cambricon's revenue soaring by 160% to reach 2.89 billion yuan. The company also experienced a 185% rise in profits, reflecting the growing demand for local computing power in China. This revenue surge comes amid rising concerns over U.S. export controls, prompting China to pursue technological self-sufficiency. Additionally, growth in the AI sector has significantly boosted demand for local chips, highlighting the importance of innovation in the Chinese semiconductor industry and its potential impact on global technology companies.

Business

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

OpenAI Fails to Meet Revenue Targets for Q1 2026

OpenAI failed to meet its internal revenue targets for Q1 2026, falling short of expected figures. This shortfall occurs amid increasing competition from major players like Google and startups such as Anthropic. A report by The Wall Street Journal noted rising tensions within the company due to substantial spending commitments, intensifying pressure on financial performance. These circumstances reflect the challenges OpenAI faces in a growing competitive landscape, signaling a need for executives in the AI sector to reassess financial and operational strategies.

Agents

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Human-in-the-Loop vs. Human-on-the-Loop: When To Use Each System

AI quality control systems can be categorized into three main approaches: human-in-the-loop (HITL), human-on-the-loop (HOTL), and hybrid systems. Each method impacts scalability, risk tolerance, and operational costs, providing diverse workflows based on task requirements. HITL involves AI performing tasks while humans retain final decision-making authority, preventing actions without approval. For instance, in loan processing, AI assesses applications but requires human approval for final decisions.

💡

I've Covered Robots for Years. This One Is Different

How might Eka's robotic claw transform our work processes?

Business

General Motors is adding Gemini to four million cars

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Business

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

General Motors is adding Gemini to four million cars

General Motors plans to integrate Google’s Gemini assistant into approximately four million vehicles in the United States. The upgrade will include Cadillac, Chevrolet, Buick, and GMC models from 2022 and onward that feature the built-in Google system. The update will be rolled out via over-the-air software updates over several months, marking one of the largest Gemini deployments in the industry. This initiative is part of GM's strategy to enhance user experience by incorporating artificial intelligence into infotainment systems.

Models

GPT-5.5 is OpenAI’s most capable agentic AI model yet

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Models

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

GPT-5.5 is OpenAI’s most capable agentic AI model yet

OpenAI launched the GPT-5.5 model on April 23, describing it as a "new class of intelligence for real work." This model is designed to excel in planning, tool usage, and output verification, enhancing performance autonomy. GPT-5.5 is the first retrained core model since GPT-4.5, developed in collaboration with NVIDIA's GB200 and GB300 systems. The practical difference is that tasks requiring multiple human interventions can now be fully delegated to the model. Results show GPT-5.5 achieved 82.7% on the Terminal-Bench 2.0 benchmark, outperforming GPT-5.4 and Claude Opus 4.7.

Models

China's 2025 AI token usage hits 21.1 quadrillion amid exponential growth

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Models

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

China's 2025 AI token usage hits 21.1 quadrillion amid exponential growth

China's cumulative AI token usage reached approximately 21.1 quadrillion by 2025, showcasing significant growth in the sector. The daily average saw a hundredfold increase within the year, indicating a substantial rise in AI adoption across various industries. This rapid growth in usage suggests a major shift in technology application in China, opening new avenues for innovation and development in the market.

Business

NTT DATA launches multivendor agentic services experience for enterprise infrastructure

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Business

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

NTT DATA launches multivendor agentic services experience for enterprise infrastructure

NTT DATA, a global leader in digital business services, has launched its Software Defined Infrastructure (SDI) Services in the Middle East and Africa. This multi-agent system enhances how enterprises manage AI infrastructure, allowing for efficient orchestration of operations. The SDI Services act as a bridge for enterprises to interact with their multi-vendor environments using natural language. This shift is crucial as organizations move from reactive operations to outcome-driven infrastructure, improving efficiency and resilience.

Agents

SenseTime Launches SenseNova U1, Moving Toward a Unified Model Era of Understanding and Generation

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Agents

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

SenseTime Launches SenseNova U1, Moving Toward a Unified Model Era of Understanding and Generation

On April 29, 2026, SenseTime officially launched the SenseNova U1 series, a unified model that integrates understanding and generation capabilities. Built on the NEO-unify architecture introduced in March, this model departs from traditional designs by eliminating separate vision encoders. It allows for efficient cross-modal collaboration between language and visual inputs.

Models

Ant Group Open-Sources Ling-2.6-Flash Model with Multiple Precision Options

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Models

29 Apr 202612 Dhuʻl-Qiʻdah 1447 AH

Ant Group Open-Sources Ling-2.6-Flash Model with Multiple Precision Options

On April 29, 2026, Ant Group announced the open-source release of Ling-2.6-Flash, featuring multiple precision formats like BF16, FP8, and INT4. The model boasts 104 billion total parameters and 7.4 billion active parameters, providing developers with flexibility across hardware environments and inference costs. Developer feedback over the past two weeks has led to optimizations enhancing bilingual switching capabilities between Chinese and English. The model's compatibility with mainstream coding frameworks has also improved, making it a strong candidate for AI applications.