Geo News
Community curated by people like you
LatestAICryptoHealthWorld AffairsUS Politics

Long-context models stories

Oct 15, 2026

Meta releases Muse Spark Open, 816B-parameter multimodal model with 1M token context

Meta launched Muse Spark Open, an open-weight AI model with 816 billion parameters that processes text, images, video, and audio with up to one million tokens of context. Apple partnered with Meta to optimize the model for Mac computers.

Oct 15, 2026·4 sources
00
Oct 12, 2026

Alibaba Qwen introduces QwenGyre elastic reinforcement learning framework for extreme-long-horizon agents

Alibaba's Qwen team has released QwenGyre, an elastic reinforcement learning framework designed to train AI agents on extreme-long-horizon tasks that run for hours and process nearly 1 million tokens. The online RL system trains coding agents through unmodified, black-box harnesses across multi-hour software engineering workloads.

Oct 12, 2026·3 sources
00
Aug 21, 2026

Ox Alpha AI model launches anonymously with 1M context window on OpenRouter

A powerful new AI model called Ox Alpha was anonymously released on OpenRouter's API platform on August 20, listed under the provider name "Stealth." The model features a 1 million token context window.

Aug 21, 2026·1 source
00
Jun 17, 2026

Z.ai Releases GLM-5.2, 753-Billion Parameter Open-Weight Model

Chinese AI startup Z.ai (formerly Zhipu AI) has released GLM-5.2, a 753-billion parameter open-weight language model that achieved the highest score (51) among open models on the Artificial Analysis Intelligence Index and reportedly outperforms GPT-5.5 on long-horizon coding benchmarks at one-sixth the cost. The model features a 1-million token context window and is available under MIT license on Hugging Face.

Jun 17, 2026·9 sources
00
Jun 1, 2026

Chinese AI Startup MiniMax Launches M3 Coding Model With Million-Token Context Window

Shanghai-based AI company MiniMax has unveiled its M3 flagship model designed for coding agents and automated workflows, featuring a one-million-token context window. The launch enters the competitive open-source AI coding space, though the company has stopped short of a full open-source commitment.

Jun 1, 2026·3 sources
00
May 7, 2026

Subquadratic Raises $29 Million Seed Round for Long-Context AI Architecture

Miami-based AI startup Subquadratic announced a $29 million seed funding round and launched SubQ 1M-Preview, described as the first fully subquadratic large language model architecture designed for efficient long-context processing.

May 7, 2026·2 sources
00
Apr 25, 2026

DeepSeek Releases V4 AI Model with Reduced Costs

Chinese AI company DeepSeek released its V4 model with drastically reduced costs and capability to process extra-long texts, though initial analysis suggests it falls short of domestic and US rivals.

Apr 25, 2026·3 sources
00
Apr 24, 2026

DeepSeek Previews V4 Model Series with Million-Token Context Windows

Chinese AI startup DeepSeek released preview versions of its V4-Pro and V4-Flash models featuring up to 1.6 trillion parameters and one-million-token context windows, claiming to close the gap with frontier models while maintaining significantly lower pricing than US competitors.

Apr 24, 2026·8 sources
00
Apr 3, 2026

Google Releases Gemma 4 Open Models Under Apache 2.0 License

Google's DeepMind released Gemma 4, a family of four open-weight AI models ranging from edge devices to workstations, marking the first time Google has used the fully open Apache 2.0 license instead of its previous restricted licensing. The models support up to 256K context windows and 140+ languages.

Apr 3, 2026·7 sources
00

Top claims

  • ▪Muse Spark Open is available free to download under a commercial license requiring the label 'Built with Muse Spark Open,' while API providers earning over $20 million annually require a separate agreement
  • ▪Meta AI chief Alexandr Wang stated that the Meta-Apple collaboration allows Meta to distribute its models widely while permitting developers to run them locally without Meta receiving prompt data or user information
  • ▪Muse Spark Open is optimized for Apple's MLX framework and can run locally across two Mac Studio computers equipped with M5 Ultra chips and 512 GB of unified memory each

Subtopics

Large language models (LLMs)6AI foundation models5AI startups4Open-source AI4AI coding assistants3AI research & benchmarks3Open weights vs closed models3AI agents2AI tools & products2AI antitrust & competition1AI compute & environmental impact1China1China AI regulations1China tech & industrial strategy1Compute, chips & AI infrastructure1Meta1Multimodal models1Muse Spark Open1Open model & data licensing1Open model families1Open models vs closed models1Ox Alpha1

Related timelines

Congress

110 stories

Crypto hacks

100 stories

Ebola outbreak

85 stories

Iran War

209 stories

Russia-Ukraine war

136 stories

Featured Stories

Z.ai Releases GLM-5.2, 753-Billion Parameter Open-Weight Model

Z.ai Releases GLM-5.2, 753-Billion Parameter Open-Weight Model

Jun 17, 2026·9 sources