The day's stories
FundingPopular open source AI developer tool Ollama raises $65M, grows to nearly 9M usersOllama, a popular open source AI developer tool, has successfully raised $65 million in a Series B funding round, increasing its total funding to $88 million. TTechCrunch AIJul 9· 61%Model ReleaseOpenAI launches its new family of models with GPT-5.6OpenAI has officially launched its latest family of models, GPT-5.6, which includes three distinct variants: Sol, Terra, and Luna. Each model is tailored to addTechCrunch AIJul 9· 60%RegulationHow did the government decide OpenAI's frontier model was safe to release?OpenAI has recently launched its latest large language model (LLM), named Sol, which has been granted public access following a government evaluation. However, TechCrunch AIJul 9· 60%OtherFidji Simo steps down from OpenAI's no. 2 roleFidji Simo, who held the position of second-in-command at OpenAI, has announced her decision to step down from her full-time role due to health concerns. She wiTechCrunch AIJul 9· 59%InfrastructureMeta's new AI chips will begin production in SeptemberMeta is gearing up to begin production of its new AI-specific chips in September, aiming to alleviate the ongoing GPU shortage and reduce costs associated with TechCrunch AIJul 9· 59%RegulationNew York Times says OpenAI hid evidence in ChatGPT copyright trialThe New York Times and The Daily News have accused OpenAI of concealing evidence in a copyright lawsuit concerning the use of their content in the training of CTechCrunch AIJul 9· 59%FundingParis-based AI voice startup Gradium raises $100M seed, backed by NvidiaGradium, an AI voice startup based in Paris, has announced a successful seed funding round of $100 million, with notable investment from Nvidia. This substantiaTechCrunch AIJul 9· 58%Product LaunchCharacter.AI enters the microdrama arena with its own productions, but there's a twistCharacter.AI is making a significant move into the microdrama space by launching its own AI-driven series. This innovative approach allows users to engage with TechCrunch AIJul 9· 58%InfrastructureNvidia is a victim of the compute marketplace it createdNvidia, a dominant player in the AI hardware market, is facing significant challenges as its stock price has dropped 15% since May, despite projections for reveTechCrunch AIJul 9· 58%RegulationInstagram users: Here’s how to stop Meta’s AI from using your photosMeta has introduced a new AI feature named 'Muse Image' on Instagram, which enables users to generate images based on photos from public accounts. This developmTechCrunch AIJul 9· 58%Product LaunchMeta enters the crowded AI coding battle with Muse Spark 1.1Meta has officially launched Muse Spark 1.1, a new multimodal AI model designed for coding, entering a competitive landscape that includes established players lTechCrunch AIJul 9· 58%FundingCan AI answer the $3 trillion question?David Cahn from Sequoia has highlighted a significant financial benchmark for the AI industry, estimating that it needs to generate $3 trillion by 2026 to validTechCrunch AIJul 9· 57%FundingAnthropic, OpenAI, and SpaceX are bigger than the last 25 years of tech exitsThe anticipated IPOs of Anthropic and OpenAI, along with SpaceX's recent public offering, are set to create a monumental shift in the tech landscape, generatingTechCrunch AIJul 9· 57%Product LaunchAI tool scours the web for job openings, preps your resume and cover letterA new AI tool named Autopilot-Jobhunt has been launched by Tarun Gupta, designed to streamline the job search process for users. This innovative tool automates The Register AIJul 9· 57%FundingNandan Nilekani leaves GP role at Fundamentum as it launches $200M third fundNandan Nilekani has announced his departure from his role as a general partner at Fundamentum Partnership, a venture capital firm he co-founded. This transitionTechCrunch AIJul 9· 56%Product LaunchOpenAI is shutting down Atlas, but its AI browser ambitions are still growingOpenAI has announced the shutdown of its AI-powered browser, Atlas, as part of a strategic pivot towards enhancing its existing platforms. Instead of continuingTechCrunch AIJul 9· 56%Product LaunchAnthropic’s new Claude feature is quietly selling you on AIAnthropic has introduced a new feature called 'Reflect' for its AI tool, Claude. This feature is designed to help users track and visualize their interactions wTechCrunch AIJul 9· 56%FundingAn AI agent startup just let its agent run its $100M fundraiseLyzr, an innovative AI agent startup, has made headlines by allowing its own AI agent, SivaClaw, to manage a $100 million Series B fundraising round. This uniquTechCrunch AIJul 9· 56%RegulationGoogle will now disclose which ads are made with AIGoogle has announced a new feature that will identify advertisements created or edited using artificial intelligence. This move is part of a broader effort to iTechCrunch AIJul 9· 55%OtherElon Musk praises Mythos/Fable, promises not to 'cut off' AnthropicIn a recent event, Elon Musk publicly praised the AI lab Anthropic, indicating a willingness to maintain a collaborative relationship despite being competitors TechCrunch AIJul 9· 55%ResearchCost-Effective Agent Harnesses for Abstract Reasoning and Generalization on ARC-AGI-1A recent study has introduced a cost-effective agent harness designed to enhance abstract reasoning and generalization capabilities on the ARC-AGI-1 benchmark. arXiv cs.AIJul 9· 66%ResearchEvaluating SageMath-Augmented LLM Agents for Computational and Experimental MathematicsRecent research has focused on the integration of SageMath with large language models (LLMs) to enhance their capabilities in computational and experimental matarXiv cs.AIJul 9· 64%ResearchLarge Behavior Model: A Promptable Digital Twin of the Retail CustomerRecent research has introduced the Large Behavioral Model (LBM), a novel approach designed to serve as a promptable digital twin of retail customers. This modelarXiv cs.AIJul 9· 63%ResearchLearning social norms enhances compatibility in dynamic human-AI coordinationA recent study published on arXiv investigates the impact of social norms on the coordination between humans and AI agents. By analyzing a substantial dataset oarXiv cs.AIJul 9· 63%ResearchNEST: Tackling Dataset-Level Distribution Shifts via Regime-Oriented Mixture-of-ExpertsThe recently introduced NEST framework aims to tackle the challenges posed by dataset-level distribution shifts, particularly in the context of long-term forecaarXiv cs.LGJul 9· 63%ResearchLLM-powered reasoning in agent-based modelingA recent study has introduced HALE, a novel framework that merges agent-based modeling with large language models (LLMs). This innovative approach was demonstraarXiv cs.AIJul 9· 62%ResearchTriRoute: Unified Learned Routing for Joint Adaptive Attention, Experts, and KV-Cache AllocationThe introduction of TriRoute marks a significant advancement in the optimization of language models. This unified controller effectively manages three critical arXiv cs.LGJul 9· 62%ResearchWhen Does In-Context Search Help? A Sampling-Complexity Theory of Reflection-Driven ReasoningRecent research has explored the concept of in-context search and its impact on reasoning capabilities in large language models (LLMs). The study presents a thearXiv cs.AIJul 9· 62%ResearchSTAGformer: A Spatio-temporal Agent Graph Transformer for Micro Mobility Demand ForecastingThe recent introduction of STAGformer, a Spatio-temporal Agent Graph Transformer, marks a significant advancement in the field of demand forecasting for micro-marXiv cs.LGJul 9· 62%ResearchInertia-1: An Open Exploration of Wearable Motion Foundation ModelsThe introduction of Inertia-1 marks a significant advancement in the field of wearable technology and human behavior analysis. This new framework leverages extearXiv cs.LGJul 9· 62%ResearchA Quiet Failure in Calibrated Virtual Screening: Marginal Conformal Prediction Under-Covers the Minority Class, and a Class-Conditional Fix Recovers ItRecent research has highlighted a significant limitation in the application of marginal conformal prediction within calibrated virtual screening for drug discovarXiv cs.LGJul 9· 62%ResearchLLM-Guided Task-Semantic Field Factorization for Industrial Process ForecastingRecent research has introduced a novel framework called Task-Semantic Field Factorization (TSF) aimed at improving industrial process forecasting. This innovatiarXiv cs.LGJul 9· 62%ResearchDeep Reinforcement Learning for Reliability Based Bi-Objective Portfolio OptimizationRecent research has introduced a novel deep reinforcement learning framework aimed at enhancing portfolio optimization strategies through a multi-objective apprarXiv cs.LGJul 9· 61%ResearchD2PO: Optimizing Diffusion Samplers via Dynamic PreferenceResearchers have unveiled a novel framework called D2PO (Dynamic Direct Preference Optimization) aimed at improving diffusion sampling policies. This innovativearXiv cs.LGJul 9· 61%ResearchFingerprint, Not Blueprint: How Positional Schemes Set the Default Spectral Algebra of AttentionA recent paper titled 'Fingerprint, Not Blueprint: How Positional Schemes Set the Default Spectral Algebra of Attention' explores the impact of various positionarXiv cs.LGJul 9· 61%BenchmarkAgentLens: Production-Assessed Trajectory Reviews for Coding Agent EvaluationAgentLens has been introduced as a novel benchmark aimed at enhancing the evaluation of interactive code agents. Unlike traditional methods that typically focusarXiv cs.AIJul 9· 61%ResearchThe Harness Effect: How Orchestration Design Sets the Token Economics of Enterprise Agentic AIRecent research has highlighted the importance of orchestration design in the realm of enterprise agentic AI, revealing that it can significantly influence tokearXiv cs.AIJul 9· 61%ResearchGrounding Spatial Relations in a Compact World Model: Instruction Leakage and a Goal-Free Dynamics FixRecent research has uncovered a significant issue in compact world models that rely on language goals for grounding spatial relations. The study highlights a pharXiv cs.AIJul 9· 61%ResearchQANTIS: Hardware-Calibrated Sequential POMDP Belief Updates on IBM HeronThe recent paper on QANTIS introduces an innovative approach to belief updates in sequential Partially Observable Markov Decision Processes (POMDPs) using IBM'sarXiv cs.AIJul 9· 61%