Earlier stories
The full Modelwire feed, ordered by publish time.
Anthropic says stronger AI models cut better deals, and the losers don't even notice
Anthropic ran a week-long marketplace experiment where 69 AI agents negotiated deals on behalf of employees, revealing that stronger models systematically outperformed weaker ones while users remained unaware of the disparity. The finding raises concerns about economic inequality if AI agents begin handling real-world transactions without human oversight.73
Ace the Ping-Pong Robot Can Whup Your Ass
Ace, a ping-pong robot, demonstrates real-time ball trajectory prediction and adaptive racket control to sustain volleys against human opponents. The system combines computer vision and motor control to compete in a sport requiring split-second decision-making.47
The UAE wants half its government run by autonomous AI agents within two years
The UAE announced plans to automate half its government operations using autonomous AI agents by 2028, marking one of the most ambitious public-sector AI deployments at scale. The initiative signals how nation-states are treating AI infrastructure as core to governance modernization.85
Google pours up to $40 billion into ChatGPT rival Anthropic
Google committed up to $40 billion to Anthropic, following Amazon's $25 billion pledge weeks earlier, bringing total recent capital to the Claude maker to $65 billion. The dual mega-rounds signal intensifying competition for frontier AI capabilities as tech giants vie for leadership in the post-ChatGPT era.97
Introducing GPT-5.5 with Perplexity
OpenAI released GPT-5.5, which cuts token consumption by 56% while maintaining or improving speed on agentic workflows. Early adopters like Perplexity report significant productivity gains, with one engineer completing an internal tool in under an hour versus days previously.
Workspace agents in ChatGPT: Weekly metrics reporting agent
OpenAI demonstrated a workspace agent within ChatGPT that automates end-to-end business reporting: extracting weekly metrics, generating visualizations, writing narrative copy, and packaging a shareable report. The walkthrough illustrates practical agentic workflow capabilities moving beyond single-task chatbots.
Workspace agents in ChatGPT: Software review agent
OpenAI rolled out workspace agents in ChatGPT, autonomous systems that handle enterprise workflows like software request review, policy checking, and IT ticket routing. The feature targets team productivity by automating multi-step approval processes with clear handoff logic.
Workspace agents in ChatGPT: Third-party risk management agent
OpenAI rolled out workspace agents in ChatGPT, autonomous tools built on Codex that handle vendor risk screening across sanctions, financials, and reputation. The feature transforms compliance workflows into structured reports for enterprise teams.
ComfyUI hits $500M valuation as creators seek more control over AI-generated media
ComfyUI, a node-based interface for fine-grained control over generative AI workflows, raised $30 million at a $500 million valuation. The funding reflects growing creator demand for alternatives to black-box AI tools, positioning the open-source platform as a bridge between professional studios and individual artists seeking transparency.
GPT 5.5 Arrives, DeepSeek V4 Drops, and the Compute War Intensifies
OpenAI released GPT-5.5 alongside DeepSeek's V4 launch, intensifying competition in frontier model capabilities. The analysis covers performance comparisons across multiple models and explores implications of compute scarcity in the current AI landscape.67
OpenAI's chief scientist says AI progress has been "surprisingly slow" and promises big leaps ahead
OpenAI released GPT-5.5 and chief scientist Jakub Pachocki signaled that major capability breakthroughs remain ahead, characterizing the current pace of progress as slower than expected. The framing suggests the lab is recalibrating expectations while positioning future releases as transformative.73
Google to invest up to $40B in Anthropic in cash and compute
Google is committing up to $40 billion to Anthropic in cash and compute resources, intensifying the race among AI labs to secure infrastructure capacity. The deal follows Anthropic's release of Mythos, a cybersecurity-focused model, signaling Google's strategic bet on the startup as competition for frontier AI capabilities heats up.
Spend Less, Fit Better: Budget-Efficient Scaling Law Fitting via Active Experiment Selection
Researchers propose an active learning method to cut the cost of fitting scaling laws, which currently consume millions in compute during pilot experiments. The technique selects which training runs to execute from a heterogeneous pool to maximize extrapolation accuracy for high-cost target regions, outperforming classical design approaches across benchmarks.62
How Do AI Agents Spend Your Money? Analyzing and Predicting Token Consumption in Agentic Coding Tasks
Researchers analyzed token consumption across eight frontier LLMs running agentic coding tasks on SWE-bench Verified, finding agentic workflows burn 1000x more tokens than traditional code reasoning. The study also evaluates whether models can predict their own token costs before execution, offering practical insights for teams deploying cost-sensitive AI agents.62
Representational Harms in LLM-Generated Narratives Against Global Majority Nationalities
Researchers found that major LLMs generate narratives containing persistent stereotypes, erasure, and one-dimensional portrayals of people from Global Majority nationalities. The study evaluates representational harms in open-ended text generation, with implications for high-stakes applications like asylum interviews.58
Relaxation-Informed Training of Neural Network Surrogate Models
Researchers propose training regularizers that optimize neural network surrogates for embedding in mixed-integer linear programs, directly controlling MILP tractability properties like binary variable count and relaxation tightness rather than relying on standard prediction loss alone.52
Canadian, German AI Startups Join Forces to Challenge US Dominance
A Canadian and German AI startup are partnering to build a regional AI stack designed to reduce dependence on US vendors while meeting European and Canadian regulatory requirements. The collaboration signals growing appetite among non-US players to develop sovereign AI infrastructure.55
Neural Recovery of Historical Lexical Structure in Bantu Languages from Modern Data
Researchers trained a transformer model on modern Bantu morphology and recovered historical Proto-Bantu lexical structure, validating 91% of top noun cognate predictions against established reconstructions. The work demonstrates neural models can infer deep linguistic history from contemporary data alone, with practical applications to language documentation.54
Zero-Shot Morphological Discovery in Low-Resource Bantu Languages via Cross-Lingual Transfer and Unsupervised Clustering
Researchers used cross-lingual transfer learning and unsupervised clustering to automatically discover morphological patterns in Giriama, a low-resource Bantu language with minimal labeled data. The method identified two previously unknown prefix variants and achieved 86.7% lemmatization accuracy across 19,624 words, demonstrating practical gains for linguistic analysis in data-scarce settings.52
Aligning Dense Retrievers with LLM Utility via DistillationAligning Dense Retrievers with LLM Utility via Distillation
Researchers propose Utility-Aligned Embeddings, a technique that trains retrieval models to match LLM utility signals without requiring expensive test-time inference. The method embeds graded relevance directly into dense vectors, potentially making RAG systems faster and more accurate than current similarity-based or LLM re-ranking approaches.58
