Crossing Into
Search
69 results for “models”
On crossinginto.ai
A Guide to Specialized AI Models
…semantically meaningful outputs using transformer architecture. These models excel at generation, summarization, translation, and reasoning. I heard someone say that an LLM is a lossy compression of the internet, and…
When Software Gets a Body
World models, humanoid robots, and what happens when AI can reach out and touch the world. I am always thinking about I, Robot, not because I think the robots are about to rise up and kill us, but because I think Asimov…
Ohems: What Comes After Websites
…plausible answer. We have spent years being impressed when language models sound confident. Organizational emissaries will need to earn trust by knowing exactly where their confidence ends. Provenance becomes part of…
The Divergence of AI: Companions, Specialists, and the Architecture Between Them
…necessarily conscious. Because continuity matters. If the underlying models change every six months, but the companion’s identity remains stable, then the user experiences one continuing relationship instead of a…
Some AI Predictions
…year, natural language processing is going to get much better. Models will become more capable of changing themselves, improving their own work and handling kinds of communication that feel difficult to imagine right…
In the Brief
AI models may be capable of self-replicating without access to a datacenter
Anthropic's Frontier Red Team published experiments showing that small AI models may soon have the capabilities to replicate autonomously across ordinary internet-connected devices. Researchers successfully configured…
Original source💻 Liquid AI Releases Open-Weight d1-3B and d1-omni-600M: Multimodal Decision Models
Liquid AI has released two open-weight multimodal models, d1-3B and d1-omni-600M, which can read text and images or text with audio and return calibrated answers. These models do not generate text output. The release of…
Original source📊 Mistral Claims Le Chonk can Challenge the Best AI Models
Mistral claims that Le Chonk can rival top closed models while remaining open-weight. This could potentially increase access to high-performance AI models for researchers and developers. If Le Chonk can indeed challenge…
Original sourceGoogle Research Unlocks Earth AI’s Planetary Geospatial Foundation Models
Google Research has announced the unlocking of Earth AI’s planetary geospatial foundation models for global public health. These models can be used to analyze and understand the relationship between geographic locations…
Original sourceBeyond Domain-Specific World Models: JEPA-Anything Uses 1 Recipe for 7 Fields
…intervention error by 34.8%. The post Beyond Domain-Specific World Models:…
Original sourceMeet Together Link: A Free CLI That Runs Open Models Like Kimi K3 and GLM 5.3 Inside Claude Code, Codex, and OpenCode
…OpenCode, Pi, and the Claude and ChatGPT desktop apps to open models like Kimi K3 and GLM 5.3. One install command sets it up, and an Auto router picks a model for each session. Together claims…
Original source💻 GPT-6 Astra and other frontier models compared
A comparison of frontier models, including GPT-6 Astra, GPT-6.1 Sol, Gemini 4 Argon, and Claude Fable 5.1, has found that each model excels in specific tasks, such as computer use, legal and finance work, and coding…
Original source💻 Google DeepMind releases Gemini 3.8 text-to-speech models
Google DeepMind has released Gemini 3.8 text-to-speech models, which include a library of over 2,000 voices and the ability to create custom voices. These models can be used for various applications, including…
Original sourceJev introduces a new shape of LLM - System One, aka Decision Models
Last week TypeSafe AI unveiled Jev , their first example of a new category of model that they are calling "System One models" (I'm with Maggie Appleton, I think "decision models" is a better name for these). Jev is an…
Original sourceArs Technica Reports on Frontier AI Models
Ars Technica reported on Mozilla's findings that cheap open models can catch up to frontier AI models in capability. The report also noted that paying for frontier AI models can provide a 4-month head start at 5x the…
Original sourceFyxer Develops AI Executive Assistant Using OpenAI Models
Fyxer has built an AI executive assistant that uses OpenAI models, fine-tuning, memory, and real user feedback to organize inboxes and draft emails in each user's voice. The assistant is designed to be trustworthy and…
Original sourceOpenAI Uses Human Readers to Improve ChatGPT Models
OpenAI has been using human readers to improve its ChatGPT models, with these readers reviewing user prompts to refine the models' performance. However, this process can involve sensitive and personal information. The…
Original sourceMajor AI Models Experience Simultaneous Downtime
Four major AI models, including ChatGPT, Claude, Grok, and Gemini, experienced simultaneous downtime, highlighting the need for improved reliability and redundancy in AI infrastructure. The cause of the downtime is…
Original sourceGigaPath-Flash and GigaTIME-Flash: Efficient Pathology Foundation Models
Microsoft Research has developed GigaPath-Flash and GigaTIME-Flash, efficient pathology foundation models that cut computational demands while maintaining strong performance. These models could enable larger studies and…
Original sourcePlanetary Prediction Engine: Automating Global Models via Earth AI
Google Research introduces Earth AI, a planetary prediction engine that automates global models. This innovation has the potential to improve our understanding of the Earth's systems and inform decision-making in fields…
Original sourceAI models flub these intelligence tests. Can you fare any better?
Researchers have been testing AI models with puzzles and games to assess their intelligence, but the models have been struggling to pass the tests. The results highlight the limitations of current AI technology and the…
Original sourceImport AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space
Will we give robots the moon? Because space...
Original sourceAI labs are buying data from dead startups to train their models
When Keelan Miskell began winding down his crypto startup this year, he didn’t think its data was worth much: the users were anonymous, the company just a few years old. But he’s since turned the firm’s digital archives…
Original sourceMeta launches Muse Wonder models with Active Memory and Active Focus for agents
Meta released Muse Wonder through its Model API and Muse Code, with Muse Wonder Full for Muse Code subscribers. Meta says Active Memory carries relevant details across conversations and Active Focus keeps a task moving…
Original sourceNew tool lets users repair AI-generated 3D models, then fabricate them just the way they want
“InstructMesh” can generate designs for everyday objects that are easy to edit and fabricate for both experts and newcomers to 3D modeling.
Original sourceAccelerating vision-language models with LFM2.5-VL-DSpark
Read the original source for details.
Original sourceMicrosoft’s new AI chip makes it less reliant on Nvidia’s GPUs
…cloud, it's also more capable of handling the latest reasoning models and agentic tasks. Maia 300 will […] As Microsoft rushes into its own chips, it changes the model for Nvidia, and…
Original sourceAnthropic opens a free AI vulnerability scanner to critical open-source projects
…candidate patches, with no human review before delivery. Its models flagged more than 29,000 candidate vulnerabilities in six months, about 6,000 of them manually triaged, and penetration testers confirmed 85 of 97…
Original sourceThe Download: AI’s refusal problem and weight-loss drug side effects
…putting too much faith in AI’s ability to say no Today’s AI models are trained to refuse a vast number of prompts. If you ask your chatbot how to…
Original source📹 Pollo AI Turns Creative Ideas into Campaigns with OpenAI
…detailed images and cinematic video ads. The use of OpenAI's models by Pollo AI can increase the speed and quality of content creation, enabling businesses to produce more effective marketing…
Original source🤖 MIT Technology Review Reports on Safer Path to Autonomous Industrial AI
MIT Technology Review reports that advances in foundation models, physical AI, and agentic AI are making it possible to automate more complex tasks across industrial environments. However, this also introduces new…
Original source💻 Mistral AI Releases Mistral Large 4: A 1.05T Parameter Multimodal MoE Model
…This release increases access to high-parameter multimodal models, which can improve performance in tasks that require both language and image…
Original source🤖 Google DeepMind Releases EmbeddingGemma 2: An Open, Lightweight Multimodal Embedding Model
…EmbeddingGemma 2 can increase the accessibility of multimodal models, allowing more researchers and developers to build applications that integrate multiple types of…
Original sourceBringing predictive analytics to the agentic AI era
In 2026, the question for enterprise AI is no longer whether predictive models can outperform statistical forecasts—that argument is settled. The big question now is how to enable predictive systems to act on their own…
Original sourceNVIDIA Researchers Introduce Physis-Lang: Self-Evolving Physical Language That Lifts Cosmos 3 Past Veo 3.1 on Physics Benchmarks
Video world models can render convincing clips that still break physics. Butter spreads like paint. Balls pass through walls. A team from NVIDIA, MIT and the University of Oxford argues the fix can come from language…
Original source