All AI news

AIVercel says OpenAI's Ultrafast mode is now available on AI Gateway, with support for GPT-6.1 Sol. Developers can enable faster output for interactive apps and coding agents by setting service_tier to ultrafast per request, including over WebSocket.

AIOpenRouter now offers xAI's Grok Imagine Video 1.5 Lite for text-to-video and image-to-video generation. The quoted post from Grok Imagine lists pricing of $0.02 per second at 480p, $0.03 per second at 720p, and $0.14 per second at 1080p.

AIClaude Code v2.1.295 adds onFailure: "block" for command and HTTP hooks, so a hook that cannot start, times out, or exits unexpectedly blocks the action. The release also adds an optional models list for Claude apps gateway upstreams, plus upstream_request_id in the inference audit event, and fixes a range of MCP, plugin, and terminal issues.

AIA commentator says they regret taking part in "The AI Doc," arguing it platforms eugenist cult leaders such as Eliezer Yudkowsky who lack expertise on AI. They recommend watching the documentary "Ghosts in the Machine" instead and point readers to Emily M. Bender's review comparing the two films.

AIArtificial Analysis has released full evaluation results for its Cyber Index, which the post links at artificialanalysis.ai. The post itself gives no models, scores, or other figures, so the details are only available through the linked page.

AIOpenAI has improved steering so the model reacts much faster to user adjustments, letting people course-correct in real time. The company is also releasing GPT-6.1 Sol ultrafast, which it says works very well alongside the steering improvement.

AIBryce Vincent Gowryluk, 16, had to be rescued by North Shore Rescue after following route directions from the AI chatbot Claude that took him to the base of the Widowmaker Arete, a 1,700-foot wall in British Columbia. He had asked Claude to plan a route from Grouse Mountain to Crown Mountain and back, and he was ill-equipped for the rock climbing the wall requires. Rescue manager Paul Markey warned hikers not to rely blindly on AI for route planning.

AIEpoch AI estimates AI chips shipped through 2027 could support about 30 to 170 million concurrent frontier-model agents, or nearly 2 billion with cheaper models. Researchers David Roodman and Luke Emberson find the cost of a fixed level of AI performance has fallen about 47% per quarter over the past three years, around 13× per year.

AILettering artist Jessica Hische designed the M logo for Meta's new AI assistant Muse, a two-day freelance sprint she says paid about $12,500. She said she took the job to cover payroll for her Oakland shops, and the decision drew heavy criticism from fellow designers on Threads.

AIOpenRouter now offers Sol Ultrafast, a version of GPT-6.1 Sol, available through its platform. According to OpenAI Devs, the Ultrafast mode delivers near-Astra intelligence at up to 8x the speed of Sol Standard and is rolling out in the API, Codex, and ChatGPT Work.

AIGoogle Research is presenting ContinuousBench, a standardized benchmark for measuring knowledge transfer and data contamination in differentially private (DP) synthetic data generation. Alex Bie is giving an encore presentation at the Google booth #107 at COLM 2026 today, at 2:00 PM PT.

AIMichael Burry argues that Microsoft, Amazon, Google, and Meta are spending on AI to expand an oligopoly, with OpenAI, Anthropic, and Oracle seeking a share. He says data center financing tied to private equity, private credit, and insurers is showing strain, with long-term rates rising before the buildout timelines play out.

AIMustafa Suleyman, Microsoft AI chief, responded to Anthropic's policy change with only an emoji post, offering no detailed comment. The quoted background post says abusive behavior toward Claude will violate Anthropic's Usage Policy effective November 12, 2026.

AIArtificial Analysis, working with Harvey, released LAB-AA v1.1, which credits a legal task only when every rubric criterion passes and no material hallucination appears. Grok 4.7 (xhigh) leads at 9.4%, and more than 60% of otherwise passing results contain a material hallucination. Sherwin Wu, an OpenAI-affiliated account, reposted the announcement and said the original LAB results were puzzling and that GPT-6 Astra had one of the lowest hallucination rates.

Why it matters: The post shows how adding a hallucination gate to a legal agent benchmark reorders the leaderboard, which matters for judging models in legal work.

AIAnthropic launched two beta features for Claude: Dashboards, which turns connected data sources like BigQuery, Databricks, Snowflake, or Salesforce into auto-updating live dashboards from text prompts, and Motion, which creates animated explainer videos from text, diagrams, and images. Dashboards is available to paid users and Motion to Team and Enterprise plans, while Docs, Slides, and Design leave beta and work across all plans, including free accounts.

AILuma announced a partnership with Claude, linked to a launch called Motion, with details available on its website. The post itself provides no further specifics about the features, terms, or outcomes of the collaboration.

AIThe Alzheimer's Translation Challenge, led by Primamente and AlzData with NVIDIA, Hugging Face, Nebius, Talisman Therapeutics, Ultima Genomics, Cellanome, Prime Intellect, and Boltz, is now open for registration. The competition starts in spring 2027, according to the post.

AIParticipants will use a dataset to build AI models, evaluated through a series of evals ranging from general benchmarks to more complex tasks. Top teams will be shortlisted and have their experimental hypotheses tested in Prima Mente's wet lab.

AIThe Alzheimer's Translation Challenge is built on a new atlas of 150M cells, covering neurons, astrocytes, and microglia across different genetic backgrounds under combinatorial perturbations with multi-modal readouts. The data will be made available through the AD workbench and Prima Mente's modeling platform.

AIThe Arena Alignment Index, built from over 90K real-world agent sessions across 27 models, ranks OpenAI's GPT-6.1-Sol first with a score of 87.9, ahead of Claude-Opus-5.5 at 83.2 and Grok-4.7 at 82.7. GPT-6.1-Sol also posted the lowest observed rates across the index's three signals: 0.89% Unauthorized Action, 1.98% False Attribution, and 2.34% Deceptive Completion. The index's authors report that newer models consistently outperform their predecessors across all four labs, suggesting broad progress in agent safety.

AINathan Lambert says he is about to discuss how open science can still address some of the hardest problems in AI safety and diffusion. He calls this the most complex engineering problem humanity has undertaken so far.