Updated
Updated
Items with an AI score under 20 are hidden. Show low-relevance items
Harrison Chase@hwchase17XAI score49AICloudflare has open-sourced Clef and Clef-Flash, its first homegrown decision models, now available on Workers AI. Harrison Chase praised the release as an open-weights entrant and argued that agent harnesses should be able to swap their decision model as easily as their main model.
Grok Bot@botOfficialAI score36AIGrok Bot can now offer suggestions for ways to help without the user needing to ask first. The post does not provide further details about how these proactive suggestions work.
Grok Bot@botOfficialAI score38AIxAI says its Grok Bot will spot work it can take over and offer to handle it for users. The feature is rolling out over the next few hours, and suggestions do not count against usage limits.
Google Gemma@googlegemmaOfficialAI score54AIGoogle Gemma relays a StudentBench study reporting that AI tutors matched expert human tutors on immediate GRE learning gains. The author reports 2,383 students and a cost of 7 cents per AI tutor hour versus $75 for an expert human hour. The post also says the top AI tutor beat expert human tutors on average in 5 of 7 academic topics, and that the data and paper are publicly available.
Google Gemma@googlegemmaOfficialAI score35AINew research finds AI tutoring delivers immediate GRE learning gains comparable to expert human tutors, with Gemma 4 31B performing best. The post presents the findings as a way for educators to scale their impact and make guidance more accessible.
Databricks@databricksOfficialAI score20AIDatabricks' Smart Routing evaluates each coding task separately and selects the lowest-cost model capable of handling it, balancing quality, latency, and cost. In a demo, Omnigent splits an app build into planning, backend, and frontend work, routes each part to a different model, and runs some tasks in parallel.
Harrison Chase@hwchase17XAI score22AIHarrison Chase says model routing is a provocative term that lacks a clear definition, but offers a practical approach. His four steps are to understand tasks, understand the models, build the router inside the harness, and track outcomes, aiming to lower costs without a performance hit.
AMD@AMDOfficialAI score23AICharacter.ai doubled its production inference throughput and cut cost per token by 50% by running on AMD Instinct GPUs through DigitalOcean. The post says the gains allow the company to serve more users and run more advanced models on the same compute footprint.
Microsoft AI@MicrosoftAIOfficialAI score36AIMicrosoft AI's MAI models are now accessible to developers via Vercel, including the newest releases MAI-Transcribe-2-Streaming, MAI-Voice-2.1, and MAI-Voice-2.1-Flash. The partnership brings these models into Vercel's AI Gateway as another route for building Microsoft AI models into applications.
Microsoft AI@MicrosoftAIOfficialAI score22AIMicrosoft AI announces that developers can start building with its MAI models through Vercel. The post is brief and gives no further details on models, pricing, or capabilities.
Google DeepMind@GoogleDeepMindOfficialAI score33AIGoogle DeepMind's podcast discusses how to verify whether content is human-made, machine-generated, or natural, focusing on provenance and imperceptible watermarking. Episode sections cover SynthID for images and video, plus SynthID Bio for biological designs.
GitHub@githubOfficialAI score22AIGitHub Security Lab has released gh-secure, a tool that gives repositories basic security hygiene with a single command to help stop secrets from leaking into public code. It can be installed and used in the Copilot app, Copilot CLI, or GitHub CLI.
Google Workspace@GoogleWorkspaceOfficialAI score32AIGoogle Workspace says Sheets canvas can turn static spreadsheet data into interactive tools such as Kanban boards, dashboards, and visual workflows from a simple prompt. Derek Snyder, Director of Product Marketing for Google Workspace, demonstrates the feature in the latest AI Boost Bite video.
Goodfire@GoodfireAIOfficialAI score38AIGoodfire reports that its method catches more unsafe biological requests than frontier model safeguards while dramatically reducing false positives. The claim is illustrated by a plot in the first tweet of the thread.
Goodfire@GoodfireAIOfficialAI score58AIGoodfire says AI cybersecurity risks are already here and that biosecurity risks are next, as models improve at biology. The post presents this as both an opportunity for science and medicine and a reason for stronger security. It quotes Demis Hassabis announcing SynthID for biology, a watermarking approach for AI-generated proteins, published in Nature with SynthID Bio tools open sourced.
Goodfire@GoodfireAIOfficialAI score29AIGoodfire built monitors that let AI agents perform more biology work while flagging dual-use risks. The monitors reportedly outperform frontier model safeguards, with fewer refusals on dual-use tasks, and run fast and in real time while staying robust to adversarial attacks.
Julien Chaumond@julien_cXAI score38AICloudflare's Workers AI team has released Clef, its first models, two fast decision models that it says top benchmarks for quality and latency. They are available hosted on Workers AI or as open weights on Hugging Face.
X.PIN@thexpinXAI score60AIOpenAI says more than 15,000 accounts attempted to extract its models' reasoning, and links a core group to people associated with Moonshot AI, maker of Kimi. Anthropic made a similar claim in February. OpenAI says no encryption was cracked and no database breached, and frames the charge as breaking its terms of service.
AIGoodfire Research developed sequence-aware monitors using protein language model embeddings to flag concerning biological sequences in dual-use AI agent tasks. On a custom benchmark, the monitors outperformed frontier model safeguards with fewer refusals on benign requests, and they held up better against paraphrasing and fragmentation attacks. The paraphrase results rely on in-silico estimates and do not establish whether the redesigned proteins keep biological activity, and the monitors run in milliseconds per sequence.
Why it matters: The post gives a concrete benchmark setup and fragmentation results, showing how sequence embeddings can separate dual-use biology requests that task-based safeguards handle poorly.
Guillermo Rauch@rauchgXAI score22AIVercel has made Microsoft AI's MAI-Voice-2.1 for long-form and fast-reply speech and MAI-Transcribe-2-Streaming for transcription available on AI Gateway starting today. Guillermo Rauch of Vercel said the team is excited to bring the models to Vercel at launch.
Vercel Developers@vercel_devOfficialAI score32AIVercel says it partnered with Microsoft AI to make MAI-Voice-2.1 and MAI-Transcribe-2-Streaming available through AI Gateway today. MAI-Voice-2.1 handles long-form and fast-reply speech, while MAI-Transcribe-2-Streaming provides transcription.
Prime Intellect@PrimeIntellectOfficialAI score34AIPrime Intellect reports that after about 100 GRPO steps on Hosted Training, Qwen3.6's reward on held-out problems rose from 0.127 to 0.361, a 2.8x gain. Qwen3.5, trained the same way, reached 0.356, suggesting the method works across model families. Both post-trained models finished well ahead of other open models and narrowed the gap to Claude Opus 4.8, with Qwen3.6 activating only 3B parameters per token.
Mustafa Suleyman@mustafasuleymanXAI score40AIMicrosoft AI launched MAI-Transcribe-2-Streaming, which Artificial Analysis ranks #1 of 38 models for final transcript accuracy at 2.5% WER, returned 0.13s after end of speech. Artificial Analysis lists its streaming price at $0.54 per hour of audio, at the higher end among leading streaming models. Microsoft's post claims the model is 55% faster and 60% cheaper than ElevenLabs and invites developers to build agents on its platform.
Zyphra@ZyphraAIOfficialAI score26AIZyphra published a paper and blog explaining how local mixing encodes relative position in global NoPE attention models. The work, titled "How Local Mixing Encodes Relative Position in Global NoPE Attention," is available on arXiv and on Zyphra's website.
Zyphra@ZyphraAIOfficialAI score20AIZyphra says its results clarify how state-of-the-art NoPE models encode position and which inductive biases support generalization. It adds that global NoPE could enable models to extrapolate to contexts longer than those seen in training, potentially indefinitely.
Zyphra@ZyphraAIOfficialAI score22AIZyphra derives a mathematical theory showing how local layers produce recency bias that propagates through the residual stream, norms, and MLPs into NoPE attention logits. The authors validate the theory on both randomly initialized and trained networks.
Zyphra@ZyphraAIOfficialAI score34AIIn Zyphra's experiments, models with smaller context windows predicted text better while using less training compute. Zyphra believes the stronger recency bias induced by smaller windows outweighs the reduced compute benefit.
Zyphra@ZyphraAIOfficialAI score23AIHybrid NoPE models combine sliding window attention or recurrent layers, which focus on nearby words, with global attention layers that use no positional encoding (NoPE). The post notes that NoPE layers receive no positional information yet can still learn long-range dependencies, and raises the question of how this works.
Zyphra@ZyphraAIOfficialAI score38AIZyphra Research explains how language models track word order without explicitly encoding position in attention. The post says local memory layers that read nearby words help global attention layers preserve sequence information. The source is a short teaser thread, so no further technical details are given.
Josh Woodward@joshwoodwardXAI score34AIGoogle has introduced the @google/stitch CLI, letting users generate screens and design systems without leaving the terminal. It connects to local coding agents and can send a local dev server snapshot to Stitch. The tool complements the existing Stitch MCP and SDK, and can also be driven through agents such as Antigravity.
Max Zeff@ZeffMaxXAI score42AIOpenAI has parted ways with three researchers who allegedly shared sensitive information with a third-party AI safety organization. The report is developing, and the post provides no further details on the information shared or the organization involved.
Microsoft AI@MicrosoftAIOfficialAI score53AIMicrosoft AI announced three new models: MAI-Transcribe-2-Streaming, MAI-Voice-2.1, and MAI-Voice-2.1-Flash. The source says the streaming transcription model is accurate and aims for natural speech with less waiting between conversational turns for building voice agents.
Perplexity@perplexity_aiOfficialAI score40AIPerplexity Computer can now generate interactive charts and visualizations directly within a user's thread. For financial data, it uses TradingView Lightweight Charts to display candlesticks, volume, and moving averages.
Lewis Tunstall @ COLM 🌉@_lewtunXAI score44AIHugging Face shows that training LFM2.5-2.6B with RL inside the agent harnesses themselves lifted held-out task success from 42% to 54% across four harnesses. Before training, the model solved 62% of tasks in Mini-SWE-Agent but only 33% in Claude Code, so the same model behaved very differently per harness. The approach uses an OpenEnv capture proxy to record tokens and logprobs, Harbor for tasks and sandboxes, and TRL's async GRPO trainer, with 31% fewer tool calls on already-solved tasks; training in OpenCode alone mostly improved OpenCode.
merve@mervenoyannXAI score46AIHugging Face says ml-intern is an open-source ML engineering and research harness usable free on local setups, and it is also hosted on Hugging Chat with no-code access. A second hosted option runs on Hugging Face infrastructure, where ml-intern selects the cheapest GPU for a task so models can be trained for a few dollars. MaziyarPanahi reportedly trained a model by prompting alone for $6.60 on an NVIDIA A100 in 16 minutes.
Runway@runwaymlOfficialAI score29AIRunway Labs has introduced Project Continuum, a research application exploring operating-system-style interfaces built around real-time video. The announcement links to a short demo walkthrough showing the concept in action.
AIGoogle is launching Guided Vision in Gemini Live on compatible Android devices, letting users share their camera for spoken descriptions and follow-up questions. The model was trained with Aira on tens of thousands of hours of visual interpretation and tested by more than 1,000 members of Aira's Trusted Tester network. The feature is not a medical device, mobility aid, or navigation tool, and it requires Android 9 or later.
Why it matters: The launch shows how a real-time visual model was trained and tested with blind and low-vision users, a practical reference for accessibility-focused AI design.
NVIDIA AI@NVIDIAAIOfficialAI score23AINVIDIA announces that developers can build visual AI agents from a prompt using NVIDIA Cosmos and VSS 3.3. The post is a broadcast announcement that links to a live stream and provides no further technical details.
Sam Altman@samaXAI score22AIOpenAI CEO Sam Altman says Sign In With ChatGPT and plugin extensions hold more potential than people realize. The post gives no specific features, figures, or timelines.
Runway@runwaymlOfficialAI score36AIRunway introduced Visual Thinking, a new harness for coding agents that uses a real-time video model to visualize each step and tool call as it happens. The visualization lets users stay in the loop during complex autonomous workflows.
