INT21 generated 20 inference engines across seven model categories in two weeks using Rust with C++ and CUDA components, and its MiMo engine reached 1,308 tokens/s versus 540 for tuned SGLang and 1,01…
Xiaomi's MiMo team launched a public dashboard at mimo.xiaomi.com/rl on September 16 that streams live training telemetry for two trillion-parameter models, MiMo-V2.6-Pro and MiMo-V2.6-Flash, with the…
Xiaomi's MiMo team is publicly livestreaming the reinforcement-learning training of two unreleased models, MiMo-V2.6-Pro and MiMo-V2.6-Flash, via a dashboard at mimo.xiaomi.com/rl that shows reward cu…
A developer investigating whether Language Server Protocol (LSP) support improves coding agent performance found mixed results. Testing the Pi agent with and without an LSP extension across three mode…
Xiaomi unveiled its flagship Xring O3 3nm AI processor for smartphones, scheduled to debut in September aboard its Xiaomi 18 Fold, alongside a 6nm Xring O100 AI accelerator and a 3nm Xring D100 chip f…
Xiaomi founder Lei Jun revealed that the company's Xring O3 chip contains a hidden 60-micron 'OK' hand gesture etched into the silicon, symbolizing 'everything is OK.' The Xring O3, unveiled on Aug. 2…
Xiaomi unveiled three TSMC-fabbed Xring chips, including the 3nm O3 flagship phone SoC, the 6nm O100 NPU for on-device MiMo LLM, and the 3nm D100 for autonomous driving, aiming to reduce reliance on Q…
An anonymous AI model dubbed Ox Alpha debuted on OpenRouter on Thursday under the stealth identifier stealth/ox-alpha, offering a 1.05-million-token context window, 131,072-token maximum output, and m…
An anonymous multimodal model named Ox Alpha appeared on OpenRouter on August 20 with a 1,048,576-token context window, multimodal reasoning across text, image, and video, and a zero-dollar price for …
A stealth model named Ox Alpha appeared on OpenRouter on August 21 under the provider name 'stealth' with no company attached, offering a 1,048,576-token context window, 131,072-token maximum output, …
A new Mac app, AI First Virtual Linux Desktop, lets AI agents run as root in a hardware-sandboxed Linux virtual machine, with no permission prompts, and includes preinstalled Claude Code, Codex, Gemin…
A developer has released 'Caged Code', a webpage that embeds the vanilla Claude Code binary and connects it to a notebook environment via WebSocket, allowing users to run Claude Code through their Ant…
AI coding platforms are shifting from unlimited plans to token-, credit-, and quota-based subscriptions. Five plans—MiniMax Token Plan, MiMo Token Plan, GLM Coding Plan, and others—offer developers fl…
A developer ran tests comparing AI companion behavior across models and relationship architectures, finding that a cheaper model with relationship architecture behaved more like a person than a strong…
Luo Fuli, head of Xiaomi's MiMo large model team, told the Beijing Academy of Artificial Intelligence Conference on June 12 that Anthropic's Claude Fable 5 should be seen as an interim-stage product, …
Xiaomi has expanded its AI portfolio across language, reasoning, code, and voice technologies, entering the open-source LLM race in April 2025 with the 7 billion-parameter MiMo model. The company's Mi…
On May 27, Xiaomi announced permanent price cuts of up to 99% for its MiMo-V2.5 API lineup, dropping cached input tokens from 2.80 yuan to 0.025 yuan per million tokens, following DeepSeek's similar 7…
DeepSeek V4 Flash defeated Xiaomi MiMo-V2.5 in a head-to-head benchmark, delivering a fully functional TypeScript implementation for room conflict detection with correct overlap detection, ISO timesta…
Open-weight models now account for 69.1% of named token volume on the OpenRouter API platform, compared to 30.9% for closed models, according to the latest platform data. The shift, driven by rapid co…
Chinese AI startup DeepSeek permanently slashed prices on its flagship V4 Pro model by 75%, undercutting Western rivals like Anthropic's Claude Sonnet and OpenAI's GPT 5.5-Med by up to 17x on outputs.…