All AI news

AIOpenAI selected AMD EPYC "Turin" CPUs to host its Jalapeño AI ASIC deployment, citing platform strength, partner experience, and reducing unnecessary risk. The post, which links to a Tom's Hardware report, says AMD aims to deliver reliable solutions for leaders facing aggressive AI performance goals.

AIllama.cpp added day-0 support for EmbeddingGemma 2, with GGUF builds available on Hugging Face under ggml-org. The post links to the embeddinggemma-2-GGUF model page as the way to run it.

AIAnthropic's Applied AI team offers virtual office hours, and Claude Startups hosts events such as Founder House, hackathons, and meetups as space allows. Startups founded in the last five years or funded in the last two are eligible to apply through the program's expansion announcement.

AIAnthropic's Claude Startup Stack combines offers from companies building with Claude, including Linear, Lovable, ElevenLabs, Granola, and Hex. Startups get discounts and credits worth up to $45,000 on tools spanning sales, design, data engineering, and more.

AIIn a Microsoft Research Podcast episode, Jennifer Neville discusses her nonlinear route into computer science and her push for more practical evaluations of today's AI systems. The post offers little beyond this framing, so no specific models, benchmarks, or results are mentioned.

AIARC Prize reports that Grok 4.7's reasoning-token usage correlates with its ARC-AGI-2 public scores. The low setting averaged about 10k tokens per test-pair attempt and scored 25%, while medium through xhigh used 86k to 120k tokens and scored 57.5% to 60%. ARC Prize suggests the lower token usage may help explain the low setting's lower score.

AIARC Prize published full Grok 4.7 results from the xAI model on its public leaderboard. The post links to the leaderboard, a GitHub repository for reproducing the results, and the ARC Prize testing policy.

AIGrok 4.7 scored 1.8% on ARC-AGI-3 in the standard harness, which lets models carry notes between turns, slightly below the 2.1% reported for Grok 4.7 in that setting. In a new provider adapter harness that preserves opaque reasoning and enables auto compaction, the score rose to 10.0%.

AIOn ARC-AGI-2, Grok 4.7's best score is 61.4% at $2.01 per task, down from Grok 4.6's 67.1% at $0.76 per task. The post notes cost per task rose 166% despite identical input and output token pricing.

AIGrok 4.7 used more reasoning tokens on average than Grok 4.6 on ARC-AGI-2 semi-private tasks at medium, high, and xhigh reasoning levels, raising its cost per task. Per test-pair attempt, medium used 136% more tokens, high 125% more, and xhigh 173% more, while low used 27% fewer. A chart compares the two models at xhigh on the 20 public tasks where Grok 4.7 increased token use the most.

AISpaceXAI's Grok 4.7 scored 90.2% on ARC-AGI-1 at $0.64 per task, higher than Grok 4.6, according to ARC Prize. It reached 61.4% on ARC-AGI-2 at $2.01 per task and 1.8% on ARC-AGI-3 under the standard harness ($2.7k), or 10.0% with the provider adapter harness ($4.8k), both lower than Grok 4.6 on those two benchmarks.

AIAi2 faculty research scientist Arman Cohan shared a thread previewing his group's upcoming COLM 2026 presentations. The background post says the work covers reinforcement learning with metacognitive rewards, on-policy self-distillation with rubric rewards, and evolving research agents. The main post itself contains only a call to see the thread, so no results or figures are reported.

AIMaarten Sap, senior research scientist and technical AI safety lead, announced that his Sapling lab has 10 papers accepted at COLM 2026. The post is a conference-acceptance announcement and includes a link to the main post, with no details on the papers' topics or findings.

AIPeter Sushko, a research engineer at Ai2, says he will attend COLM and that his team is hiring senior engineers, pre-docs, and interns. The post is a recruiting announcement sharing the team's open roles.

AIAi2 applied scientist Shashank Gupta says he will attend COLM 2026 from Tuesday through Friday and invites people to talk with the Ai2 Agents post-training team. Topics include post-training for coding and long-horizon agents, such as agentic RL, OPD, data and infrastructure, and multi-agent training, plus opportunities at Ai2. He lists an Ai2 booth session Tuesday 1:30–3pm and an Ai2 mixer Tuesday 6–9pm.

AIKevin Farhat, a research engineer at Ai2, says he is attending COLM all week and invites people to hang out. Ai2 is also hosting an event for attendees to meet the team, learn what it is building, and explore open roles.

AIAi2 announced William Merrill as a Young Investigator, linking to a post featuring his work. The post itself offers no details on his research or the award's specifics. A quoted post notes Merrill will attend COLM in San Francisco from Tuesday through Sunday.

AIPao Siangliulue — Research Scientist 👇

AIGoogle's EmbeddingGemma 2 is an open embeddings model for on-device use that covers code, image, video, audio, and text. It comes in modular sizes from 270M text/code to 740M full multimodal, supports Matryoshka truncation down to 128 dimensions, and reports a 14% gain on MTEB Code over v1 under an Apache 2.0 license. The author's post highlights the release and a Hugging Face demo, while the benchmark table compares it with several models.

AILauren Tan describes using Grok Bot to connect apps like Slack, email, and CRM so it can gather context and decide what work to do. She also explains scheduled and event-triggered routines, plus handing engineering tasks to Cursor cloud agents, with a daily feedback-to-merge loop example.

AIKrea (@krea_ai) shared a link to a Nano Banana 2.1 image page on its website. The post provides no further details about the model's capabilities, features, or pricing.

AIKrea has made Nano Banana 2.1 available on its platform, promising better prompt understanding, sharper edits, and stronger subject consistency. The model now supports generations up to 4K resolution.

AIAfter a recent ride in Waymo's Ojai vehicle, NVIDIA's Bryan Catanzaro agreed that the driving was jerky. He said braking was rougher even in easy situations like pulling over at the end of the ride. The interior was roomy, but he judged the ride quality poor.

AIThe author spent two days building an internal enterprise platform on all Feishu data and a self-built MCP, replacing Feishu's native workbench to handle Vibe Coding app deployment, security, permissions, and app and skill circulation. A custom configuration interface is planned so employees can use their own Agents to modify their homepages and data pages.

AIAmpersand lets developers connect an AI agent to a customer's Salesforce by configuring object and field mappings. The platform then handles API calls, authentication, token refreshes, and retries, which the post presents as a major advantage given the difficulty of managing multiple Salesforce accounts.

AIAt the Machine Earning AI Summit in San Francisco, Town co-founder Jean-Denis Greze said personal AI agent purchases will initially require human approval, with mistakes budgeted in like credit card fraud. Lead Bank CEO Jackie Reses raised liability questions, asking "If a model hallucinates, whose responsibility is that?" Chime co-founder Ryan King said AI agents will make switching banks easier, though regular people are not yet ready to let AI manage their money.

AIMicrosoft highlights AT&T data scientist Natalie Gilbert, who applies skepticism to AI by examining edge cases and making informed decisions rather than accepting AI outputs blindly. The post frames questioning as the basis for trustworthy AI, linking to a Microsoft feature titled "See you in the work."

AIMeta AI leader Alexandr Wang reacted to The Daily Show's segment featuring Jon Stewart talking with Meta's Muse AI agent, asking which corporate figure booked the appearance. The show's background description portrays Muse as an AI agent Stewart chats with in a joking, suspicious tone.

AIJoshua Achiam argues that many in Silicon Valley wrongly define success as controlling the largest share of matter and energy in the universe, a goal beyond human limits that can drive them toward successionism. He contends that success instead comes from inner lives, relationships, creativity, cooperation, and striving to overcome human limitations, which could make them less pessimistic.

AI跟着东锡老师学关键词： Sovereign AI

AIGoogle's Nano Banana 2.1, model gemini-nano-banana-2.1, is now available and is said to outperform the previous Pro model at about a quarter of the price, $0.034 per image versus $0.134. The quoted post lists improved instruction following, better in-image text rendering, grounding with Google Image Search, and up to 5 characters of consistency plus 14 reference images. It is available in Google AI Studio, the Gemini API, Google Cloud, the Gemini app, and Flow. The author's own post is a playful reaction praising its design ability and shows a generated vegan basketball food truck poster.

AIMicrosoft Research podcast host Chad Atalla interviews Jennifer Neville, a partner research manager at Microsoft, about her path into AI and her work on how evaluation exposes surprising failures in models tested beyond traditional benchmarks. The conversation also covers practical guidance for working with current AI systems and why examining underlying data matters when results defy expectations.

AIGoogle released EmbeddingGemma 2, a 740M-parameter open embedding model under Apache 2.0 that combines a 270M text model with vision (170M) and audio (300M) encoders. The 270M text model can run locally with 0.5GB of RAM, and the full multimodal model with 1GB, and Unsloth provides GGUF files and fine-tuning support.

AIGoogle has made EmbeddingGemma 2 available to download on Hugging Face and Kaggle. Availability in the Gemini Enterprise Agent Platform Model Garden is coming soon.

AIGoogle's EmbeddingGemma 2 is a 740M-parameter embedding model that outperforms some models more than twice its size while using about 191MB to 567MB of active RAM. It offers an 8K context window, 4x larger than the first generation, and can process up to 5.5 minutes of audio, 29 images, or 58 video frames in one pass.

AIGoogle says EmbeddingGemma 2, paired with Gemma 4, enables efficient on-device retrieval-augmented generation with a lower memory footprint. In this setup, EmbeddingGemma 2 retrieves local files and Gemma 4 reasons over them to produce grounded answers while keeping data private.

AIThariq argues that working at a higher level of abstraction has always required understanding the levels beneath it. He says coding agents do not change this requirement.

AICodex now offers an "Approve for me" permission mode, which automatically reviews actions instead of requiring manual approval. To enable it, open the permissions menu below the composer and select "Approve for me."

AISuno announced a music-making event with students at Seoul Institute of the Arts. The post gives no further details about the program, participants, or outcomes.

AIAccording to Gergely Orosz, GitHub now records more AI-generated pull requests per month than human-written ones. He says the volume of AI-generated PRs is about three times what it was in December 2023.