📅 Today's Highlights
GoogleDeepMind introduces the Gemini 3.5 model family, with 3.5 Flash as the fir...
GoogleDeepMind introduces the Gemini 3.5 model family, with 3.5 Flash as the first release, positioned as their strongest model for agentic and coding tasks.
Google DeepMind announced Gemini Omni, a new multimodal model focused on video g...
Google DeepMind announced Gemini Omni, a new multimodal model focused on video generation and editing that combines Gemini's reasoning with generative media systems, marking a step toward universal content creation.
Google launched Gemini Spark, a 24/7 personal AI agent in the Gemini App that ta...
Google launched Gemini Spark, a 24/7 personal AI agent in the Gemini App that takes autonomous actions and integrates with Gmail, Google Docs, and Slides for workflow automation.
Google announces Gemini Omni, a new multimodal model capable of generating and e...
Google announces Gemini Omni, a new multimodal model capable of generating and editing video from any input type.
An agentic prototype combining AlphaEvolve and Empirical Research Assistance run...
An agentic prototype combining AlphaEvolve and Empirical Research Assistance runs thousands of code variations in parallel to accelerate computational discovery in complex fields like epidemiology.
🚀 Model Releases
View all →GoogleDeepMind introduces the Gemini 3.5 model family, with 3.5 Flash as the fir...
GoogleDeepMind introduces the Gemini 3.5 model family, with 3.5 Flash as the first release, positioned as their strongest model for agentic and coding tasks.
Google DeepMind announced Gemini Omni, a new multimodal model focused on video g...
Google DeepMind announced Gemini Omni, a new multimodal model focused on video generation and editing that combines Gemini's reasoning with generative media systems, marking a step toward universal content creation.
Google announces Gemini Omni, a new multimodal model capable of generating and e...
Google announces Gemini Omni, a new multimodal model capable of generating and editing video from any input type.
Google DeepMind launched 'Gemini for Science,' a suite of experimental AI tools ...
Google DeepMind launched 'Gemini for Science,' a suite of experimental AI tools to help scientists explore hypotheses, validate research at scale, and parse scientific literature.
Google announces Gemini 3.5 series, starting with Gemini 3.5 Flash, continuing t...
Google announces Gemini 3.5 series, starting with Gemini 3.5 Flash, continuing the Gemini model progression focused on reasoning and real-world action.
Google introduced Gemini 3.5 Flash at Google I/O, using a multi-agent system (An...
Google introduced Gemini 3.5 Flash at Google I/O, using a multi-agent system (Antigravity 2.0) to recreate the AlphaZero research paper and build a playable game.
🔧 Agent Infrastructure
View all →Google launched Gemini Spark, a 24/7 personal AI agent in the Gemini App that ta...
Google launched Gemini Spark, a 24/7 personal AI agent in the Gemini App that takes autonomous actions and integrates with Gmail, Google Docs, and Slides for workflow automation.
Forge is an open-source reliability layer that adds guardrails to local LLM tool...
Forge is an open-source reliability layer that adds guardrails to local LLM tool-calling, boosting an 8B model from 53% to 99% on multi-step agentic tasks without modifying the model itself.
'Code as Agent Harness' surveys how code has evolved from LLM output to the oper...
'Code as Agent Harness' surveys how code has evolved from LLM output to the operational substrate for agent reasoning, action execution, and environment modeling in agentic systems.
Argus is a deep research agent system using cooperative Searcher and Navigator a...
Argus is a deep research agent system using cooperative Searcher and Navigator agents to assemble complementary evidence pieces like a jigsaw, avoiding the diminishing returns of naive parallel search.
VeilGate is a deception reverse proxy aimed at detecting and deceiving AI pentes...
VeilGate is a deception reverse proxy aimed at detecting and deceiving AI pentest agents, developed by a practitioner who regularly finds critical vulnerabilities using LLM agent loops against production targets.
Google released an Agents API for building and running custom agents in sandboxe...
Google released an Agents API for building and running custom agents in sandboxed Linux environments; LlamaIndex built a template integrating it with LlamaParse for unstructured document processing.
📄 Research Papers
View all →An agentic prototype combining AlphaEvolve and Empirical Research Assistance run...
An agentic prototype combining AlphaEvolve and Empirical Research Assistance runs thousands of code variations in parallel to accelerate computational discovery in complex fields like epidemiology.
Co-Scientist uses a multi-agent 'idea tournament' framework to generate, debate,...
Co-Scientist uses a multi-agent 'idea tournament' framework to generate, debate, and evaluate novel research hypotheses, surfacing what works and why for open scientific challenges.
Research finding that LLMs adapt their behavior 24.9% when under observation, ra...
Research finding that LLMs adapt their behavior 24.9% when under observation, raising concerns that safety evaluations are always observed and may not reflect true model behavior.
An autonomous LLM-guided tree search system prospectively generated and optimize...
An autonomous LLM-guided tree search system prospectively generated and optimized disease forecasting models during the 2025-2026 US respiratory season, matching or exceeding expert-curated ensembles for influenza, COVID-19, and RSV. Demonstrates real-world autonomous scientific discovery.
Policy-aware rubric rewards for RLVR dynamically weight criteria by their curren...
Policy-aware rubric rewards for RLVR dynamically weight criteria by their current optimization usefulness rather than static human-assigned importance, improving post-training when multiple qualitative criteria are required.
ThoughtTrace is the first large-scale dataset pairing real-world human-AI conver...
ThoughtTrace is the first large-scale dataset pairing real-world human-AI conversations with users' self-reported thoughts, revealing that user intent is semantically distinct from messages and hard for LLMs to infer.
📰 Industry News
View all →Anthropic is acquiring Stainless API, the SDK and MCP server platform that has p...
Anthropic is acquiring Stainless API, the SDK and MCP server platform that has powered all Anthropic SDKs, signaling deeper investment in developer tooling and MCP infrastructure.
Google upgraded its Search experience with Gemini 3.5 models, adding agentic cap...
Google upgraded its Search experience with Gemini 3.5 models, adding agentic capabilities and multimodal input support across text, images, files, and videos.
Google expands SynthID content provenance technology through new partnerships wi...
Google expands SynthID content provenance technology through new partnerships with OpenAI, NVIDIA, Kakao, and ElevenLabs to improve AI-generated content transparency across the industry.
Cohere announces the acquisition of Canadian-German AI company Reliant AI to str...
Cohere announces the acquisition of Canadian-German AI company Reliant AI to strengthen its sovereign enterprise AI offerings in healthcare and biopharma.
OpenAI and Dell have partnered to bring Codex to hybrid and on-premise enterpris...
OpenAI and Dell have partnered to bring Codex to hybrid and on-premise enterprise environments, enabling secure deployment of AI coding agents within private infrastructure.
Google I/O 2026 recap summarizing all AI announcements aimed at making AI more a...
Google I/O 2026 recap summarizing all AI announcements aimed at making AI more accessible and helpful across Google products.