MiniMax Launches Open-Weight M3 Model to Challenge Closed-Source Leaders
Chinese AI firm MiniMax has unveiled M3, an open-weight model with a 1-million-token context window and native multimodal input. The company claims M3 surpasses GPT-5.5 and Gemini 3.1 Pro on key coding benchmarks while running at a fraction of the cost, though independent verification is still pending.
Microsoft unveils in-house AI models to cut costs and reduce OpenAI reliance
At its Build 2026 conference, Microsoft introduced the MAI family of fully in-house AI models. With the coding model MAI-Code-1-Flash and the reasoning model MAI-Thinking-1, the company aims to lower developer costs while reducing its dependence on OpenAI.
Nvidia enters the laptop market with new RTX Spark superchip
Nvidia unveiled its Arm-based RTX Spark superchip at Computex 2026. With 128GB of unified memory and a Blackwell GPU, it will power more than 30 laptops from Dell, HP, Lenovo and Microsoft Surface starting this fall.
OpenAI Launches Three New Realtime Voice Models With GPT-5-Class Reasoning
OpenAI on May 7, 2026 released three new realtime audio models — GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper — through its Realtime API. The lineup brings GPT-5-class reasoning to voice, live translation across 70+ languages, and low-latency streaming transcription.
Subquadratic Launches SubQ With 12-Million-Token Context Window and $29M Seed Round
Miami-based AI startup Subquadratic has emerged from stealth with SubQ, a large language model offering a 12-million-token context window, alongside a $29 million seed round. The company says its new sparse-attention architecture scales linearly with context length, sharply lowering long-context compute costs.
DeepSeek Launches Open-Weight V4 With 1M-Token Context and Aggressive Pricing
Chinese AI lab DeepSeek has released its V4 preview with two open-weight Mixture-of-Experts models: the 1.6-trillion-parameter Pro and the 284-billion-parameter Flash. Both support a 1-million-token context window under the MIT license, and pricing undercuts GPT-5.5 and Claude Opus 4.7 by a wide margin.
OpenAI Launches ChatGPT Images 2.0 With Built-In Reasoning and 2K Output
OpenAI has launched ChatGPT Images 2.0, a new image model with native reasoning, 2K output, and stronger multilingual text rendering. Released April 21, 2026 as gpt-image-2, it is available across ChatGPT, Codex, and the API.
Anthropic Launches Claude Opus 4.7, Outperforming GPT-5.4 and Gemini 3.1 Pro Across Key Benchmarks
Anthropic has released Claude Opus 4.7, its most capable publicly available model to date. The upgrade delivers double-digit gains in agentic coding, vision, and instruction-following, surpassing GPT-5.4 and Gemini 3.1 Pro on most major benchmarks while maintaining the same pricing as its predecessor.