Xiaomi has released MiMo-V2.5-Pro-UltraSpeed: The generation speed has been increased by 10 times! It can process over 1,000 tokens per second.

igeekphone.com Jun 9, 2026

Xiaomi has released MiMo-V2.5-Pro-UltraSpeed: The generation speed has been increased by 10 times! It can process over 1,000 tokens per second.

Curated by Shubham · TechCapsules · Jun 9, 2026

AI-summarised brief · reviewed before publication

Xiaomi has launched MiMo-V2.5-Pro-UltraSpeed, a breakthrough in AI text generation speed, achieving 1000 tokens per second on a single standard 8-card GPU node. This is a 10-fold increase in speed, made possible by collaborations with TileRT and innovations in model and system design. The service is available on a limited-time subscription basis, with priority given to enterprises and professional developers. A dedicated webpage allows ordinary users to experience the conversation function, with a daily queue limit and session duration. The technology advancements include FP4 quantization, DFlash block parallel speculative decoding, and a restructured GPU execution architecture.

💡 Why It Matters

· Faster AI text generation enables parallel reasoning of models and autonomous error correction, significantly improving logical reasoning quality.
· This boosts productivity in code generation and programming agents.

Capsules LLMs

Read full article on igeekphone.com

Elon Musk predicts Grok will start to challenge Hollywood by the end…

Elon Musk has announced that xAI's model, Grok, will be capable of creating full movies by the end of 2026. A trailer for Homer's The Odyssey, created using Grok Imagine Video 1.5, showcases the AI's rapid strides in video generation. The 2-minute-plus trailer features 36 consistent shots, realistic special effects, and emotional depth, demonstrating Grok's capabilities in scale, physics, and cinematic workflow. Musk believes Grok will be "watchable" by the end of 2026 and "really good" by 2027.

💡 Hollywood's dominance may be challenged by AI-generated content, as Grok's capabilities raise questions about the future of cinematic authorship.

techtimes.com · Jun 18

Satellite AI Inference Clears Orbit: Gemma 3 Ran Aboard YAM-9 in April

Loft Orbital's YAM-9 spacecraft successfully demonstrated the first publicly disclosed execution of a vision-language model in orbit. Aboard the spacecraft, Google DeepMind's Gemma 3 performed natural language queries, classified Earth imagery, and produced plain-English summaries without transmitting raw pixels to the ground. This achievement marks a significant shift in the value layer of Earth observation, moving from ground-based processing to onboard inference capabilities. The model, running on an Nvidia Jetson Orin AGX graphics processing unit, utilized a software harness developed by NASA's Jet Propulsion Laboratory to fit within the spacecraft's memory budget and processing constraints.

💡 This breakthrough enables Earth observation satellites to operate more autonomously, reducing reliance on ground-based processing and potentially leading to faster and more efficient analysis of satellite imagery.

interestingengineering.com · Jun 18

China’s Alibaba unveils AI brains designed to power the next generation of…

Chinese firm Alibaba has unveiled its first embodied AI model family, the Qwen-Robot suite, which links large language models with real-world robotic actions. The suite comprises three models focused on navigation, manipulation, and world modeling for robots operating in physical environments. Developed by Alibaba's Tongyi Lab, the Qwen-Robot suite enables machines to perceive, reason, and interact with the real world, joining a growing global push to advance embodied AI beyond traditional chatbot applications. The models are currently undergoing pilot testing with selected Alibaba Cloud enterprise clients.

💡 The Qwen-Robot suite's success could accelerate the development of robots capable of performing complex tasks in real-world environments, potentially transforming industries such as manufacturing, logistics, and healthcare.

thestandard.com.hk · Jun 17

US holds off blacklisting China’s DeepSeek, more than 100 firms deemed security…

The US has held off adding China's AI startup DeepSeek and over 100 other companies to a trade blacklist, despite being flagged as national security risks. DeepSeek, which has supported China's military and intelligence operations, and memory chipmaker CXMT, were approved for addition to the Entity List last year. The list, which restricts US companies from shipping goods to listed companies without a license, has not been updated since October. The delay is seen as an effort to avoid escalating tensions with Beijing, amidst a tense rivalry over technology, trade, and national security. The Entity List is a key tool in combating bad actors.

💡 Delaying the addition of these companies to the Entity List may allow American technology to reach adversaries, potentially undermining national security. The lack of updates to the list also undermines the effectiveness of trade policy tools.

techtimes.com · Jun 17

Grok V9-Medium Arrives as SpaceX Seals Cursor: Developers Face Model-Choice Risk

SpaceX has signed a $60 billion merger agreement to acquire Anysphere, the company behind AI coding tool Cursor, following its record Nasdaq debut. The deal coincides with the public release of Grok V9-Medium, a 1.5-trillion-parameter coding model trained on Cursor developer workflows. This raises questions for over one million developers who use Cursor, which currently runs on multiple models. Grok V9-Medium is optimized for NVIDIA's Blackwell GPU architecture and is designed to close the coding performance gap against other models. The model's training data includes real Cursor developer sessions, setting it apart from other models.

💡 SpaceX's acquisition of Cursor gives it full ownership of valuable workflow data, potentially altering the competitive landscape. The integration of this data into Grok V9-Medium could give SpaceX a unique advantage in the coding model market.

techradar.com · Jun 17

People are starting to think ChatGPT is too cheap — and that…

A growing number of ChatGPT users are questioning the value of the service, wondering if they're getting away with something due to its low price. The $20-a-month ChatGPT Plus and $200-a-month ChatGPT Pro subscription prices haven't changed since their introduction, despite the service becoming more powerful and feature-rich. OpenAI CEO Sam Altman previously stated that ChatGPT Pro loses money for the company due to its popularity. The concern arises from the significant computing resources required to power the AI system, with vast networks of specialized hardware and data centers consuming large amounts of electricity. As ChatGPT's capabilities expand, users are increasingly asking if it costs less than it should, sparking debate about its pricing.

💡 The perceived undervaluing of ChatGPT threatens OpenAI's ability to sustain its service, as the company invests heavily in data centers, hardware, and electricity to support its growing user base. AI's high operational costs make it vulnerable to pricing pressures, potentially disrupting the long-term viability of ChatGPT and similar services.