Google DeepMind has introduced Gemini 4 Argon, its new frontier model featuring an industry-leading 1-million-token output limit and advanced capabilities for software engineering and cyber defense.
This brief is compiled automatically by AI and may contain errors. Check the linked sources.
Model Releases & Updates
Gemini 4 ArgonGGoogle DeepMind
Bottom line：Google DeepMind has introduced Gemini 4 Argon, its new frontier model featuring an industry-leading 1-million-token output limit and advanced capabilities for software engineering and cyber defense.
Output Capacity：Supports up to 1 million output tokens in a single response, allowing models to tackle long, multi-step workflows without splitting tasks.
Availability：Initially rolling out to trusted cyber defenders through the Fairwind Program before expanding to developers and consumers.
Pricing：Priced at launch at $2 per million input tokens and $10 per million output tokens, with cached input discounted to 5% of standard rates.
Bottom line：InclusionAI has released Ling 3.1 Flash, a hybrid reasoning language model featuring 560 billion total parameters and a 262K-token context window.
Architecture：Employs a hybrid linear architecture with an increased linear Attention layer ratio and 25B active parameters per token.
Capabilities：Optimized for long-document processing, multi-step analytical reasoning, coding, and tool-use agent workflows.
Bottom line：Cloudflare has rearchitected its Containers platform to deliver 6x faster startup times and optimized support for on-demand agent workloads.
Performance：Median startup times dropped from over four seconds down to 648 milliseconds, enabling rapid burst testing.
Programmability：Allows application code to dynamically choose sandbox images and instance types at runtime with filesystem snapshots in public beta.
Bottom line：Inworld has acquired Ultravox, uniting an advanced conversational voice model research lab with a developer platform for real-time voice agents.
Integration：Combines Inworld’s proprietary voice intelligence with Ultravox’s real-time runtime infrastructure for building voice agents.
Capabilities：Enables developers to build real-time voice agents that handle natural interruptions, pauses, and complex tool-calling workflows.
Bottom line：Salesforce has acquired AI startup Listen Labs for approximately $2 billion following a rapid growth trajectory.
Deal Scope：The transaction values Listen Labs at roughly $2 billion less than three years after its founding and with under $100 million in total funding.
Founders：The founding team turned down a prior unicorn-level term sheet to pursue the strategic acquisition.
Bottom line：DeepSeek has officially open-sourced its core infrastructure components for Huawei’s Ascend compute platform, aligning with its NVIDIA ecosystem tools.
Components：Includes the TileLang programming language, DeepGEMM matrix acceleration, DeepEP communication library, and FlashMLA sparse attention.
Collaboration：Developed in close cooperation with Huawei to support Ascend 950-based 128-card supernode architectures with optimized compute and communication.
Ataraxos: Mastering StrategoMMIT, CMU, NYU & Stanford
Bottom line：Researchers from MIT, Carnegie Mellon, NYU, and Stanford have developed Ataraxos, an AI system that surpasses elite human players in the imperfect-information game Stratego.
Milestone：Achieved significant outperformance against world-class human experts in tactical and hidden-information reasoning.
Publication：The research findings were published in the scientific journal Nature.
