Search the archive

OPEN SOURCE OPENAI PRODUCTION KERNELS (GITHUB PR)

OpenAI has open sourced some of its fp4 and MoE kernels to the Triton language GitHub.

NEMOTRON H MODELS (HUGGING FACE HUB)

Nvidia's ADLR team has released the weights for the hybrid mamba Nemotron models, which have strong long context performance and decent general benchmark abilities.

AUTO DEPLOY

A new way to transform PyTorch and Hugging Face models into a faster, deployable, format for fast inference.

OPENAI'S UPDATES PREPAREDNESS FRAMEWORK

OpenAI has revised its safety strategy and introduced sharper risk categories, stronger evaluation processes, and new guidelines for managing advanced AI capabilities.

CANADA HAS ANSWER TO ENERGY NEEDS IN AI RACE, EX-GOOGLE CEO SAYS

Canada has 250 data centers and aims to expand by leveraging its abundant energy resources, like hydro and nuclear. Conservative leader Pierre Poilievre advocates for increased resource production to boost economic benefits domestically.

JACK MA ADVOCATES FOR AI TO SERVE HUMANITY, NOT DOMINATE

Jack Ma advocates for AI development focused on enhancing human life, warning against technologies that dominate. He urges global collaboration to create ethical guidelines, ensuring AI complements rather than replaces human capabilities. His call reflects broader public sentiment favoring technology that supports societal welfare.

your spot for the webinar

NOTION LAUNCHES AI-POWERED EMAIL

Notion Mail is a Gmail-integrated client that helps users manage, search, and reply to email using AI.

OPENAI SLASHES AI MODEL SAFETY TESTING TIME

OpenAI has reduced the time and resources for safety testing its AI models, raising concerns over rushed releases without proper safeguards.

CHATGPT BECAME THE MOST DOWNLOADED APP GLOBALLY IN MARCH

ChatGPT became the most downloaded non-gaming app in March, surpassing Instagram and TikTok with 46 million downloads.

OPENAI GPT-4.1

OpenAI has launched three new models in its API: GPT‑4.1, GPT‑4.1 mini, and GPT‑4.1 nano. These models outperform GPT‑4o and GPT‑4o mini across the board, with major gains in coding and instruction following. They also have larger context windows—supporting up to 1 million tokens of context—and are able to better use that context with improved long-context comprehension. They feature a refreshed knowledge cutoff of June 2024.

HUGGING FACE ACQUIRES POLLEN ROBOTICS

Hugging Face, the center of the open source AI community, has long stated its goal is to be a decentralized DeepMind. While this isn't exactly the case, adding in an open source robotics platform via Pollen moves it closer to that goal.

DOLPHINGEMMA

DeepMind has announced DolphinGemma, a large language model developed by Google that helps scientists study how dolphins communicate — and hopefully find out what they're saying, too.

SEAWEED-7B: COST-EFFECTIVE TRAINING OF VIDEO GENERATION FOUNDATION MODEL

The ByteDance team has released a paper showing how to train a competitive 7B parameter video generation model on a “modest” compute budget of 655k H100 hours. It has strong performance on a number of temporally difficult tasks.

PIXELFLOW: PIXEL-SPACE GENERATIVE MODELS WITH FLOW

Most generative models on continuous signals operate in latent space due to computational constraints. This work introduces a series of cascades that allow the generation to happen directly in pixel space. This eliminates the need for a pretrained VAE.

INTERACTVLM: 3D INTERACTION REASONING FROM 2D FOUNDATIONAL MODELS

New VLM that can reason about contacts between humans in 3D and objects. It does so by leveraging a strong base model and lifting its reasoning into 3D with clever multi-view rendering.

3B PARAMETER TOKENIZER

Scaling up image tokenizers is challenging because they tend to collapse. This work introduces GigaTok, which is a massive tokenizer with superior reconstruction performance. Decoder scaling and regularization helped with stability and overall quality.

IMPROVED MOE WITH C3PO

C3PO introduces a new test-time optimization technique that improves accuracy in Mixture-of-Experts LLMs by re-mixing expert weights based on similar reference samples.

VISUAL REASONING WITH LESS DATA

Using MCTS to quantify sample difficulty, ThinkLite-VL improves reasoning in VLMs with just 11k training samples and no distillation.

BROWSECOMP BENCHMARK FOR HARD-TO-FIND KNOWLEDGE

OpenAI's BrowseComp is a new benchmark of 1,266 problems designed to evaluate AI agents' browsing skills in gathering complex, hard-to-locate information online.

BUSINESS LEADERS' THOUGHTS ON AI POSSIBILITIES

Executives from nine companies share how they're leveraging Google Cloud's AI tools to drive innovation across sectors, with over 600 real-world use cases highlighted.

6 HIGHLIGHTS FROM GOOGLE CLOUD NEXT 25

Vertex AI introduces updates to video, image, speech, and music generation models, enhancing creative workflows for businesses. Google AI is enabling specialized AI agents for companies, improving productivity and security. A new Agent2Agent Protocol allows different AI agents to securely communicate across platforms.

DEEPSEEK TO OPEN SOURCE ITS INFERENCE ENGINE

DeepSeek's inference engine is built on VLLM, although it is now heavily modified.

NVIDIA TO MANUFACTURE AI SUPERCOMPUTERS IN THE U.S.

NVIDIA is localizing AI hardware production by building factories in Texas and Arizona, aiming to produce Blackwell chips and AI supercomputers entirely within the U.S.

GEMINI ADDS QUESTION GENERATION TO GOOGLE CLASSROOM

Educators can now use Gemini to generate questions or quizzes from selected text in Google Classroom, enhancing lesson interactivity and streamlining content creation.

SAFE SUPERINTELLIGENCE VALUED AT $32B

OpenAI co-founder Ilya Sutskever's AGI moonshot was recently valued at a substantial leap from its September funding round.

AIRTABLE'S NEW AI ASSISTANT

Airtable has launched Airtable Assistant in beta alongside new capabilities that drastically change what it means to build apps in Airtable. The updates make it easier than ever for developers to build and modify apps, create automations and interfaces, and answer any business questions on enterprise data using natural language.

VERIFIED ID MAY BE REQUIRED FOR FUTURE OPENAI API ACCESS

OpenAI plans to gate access to certain upcoming models behind a new Verified Organization process that will require government-issued ID. Verification will be limited to one org per ID every 90 days.

ANTHROPIC REPORT ON UNIVERSITY STUDENT AI USAGE

3D OBJECT PART SEGMENTATION

HoloPart is a semantic 3D Object segmentation model that can identify and separate a single 3D object into meaningful sub-pieces.

EFFICIENT COMPRESSION FOR MOE MODELS

C-Prune is a two-stage pruning method that compresses Mixture-of-Experts models by clustering similar experts and pruning redundant clusters.

JAX RECOMMENDATION ENGINE

A great recommendation engine with metrics, implementations of embedding models, and training infrastructure.

REASONING VLM FROM KIMI

One of the first open reasoning models for visual question answering. This small model is extremely good at grounded question answering for images, image captioning, and even some math based on image input.

DOCUMENT RERANKING

LLM4Ranking is a new modular framework that supports both open and closed LLMs for document reranking, complete with evaluation tools and reproducible benchmarks across popular datasets.

GOOGLE LAUNCHES AI SHORT FILM INITIATIVE

Google has partnered with filmmakers to produce AI-themed short films, aiming to explore the emotional and ethical dynamics of human-AI coexistence.

YOUTUBE BACKS NO FAKES ACT TO FIGHT AI MISUSE

YouTube supports new legislation to combat AI-generated impersonations, reinforcing its commitment to protecting creators and viewers from deepfake harms.

CANVA UNVEILS VISUAL SUITE 2.0 WITH AI-POWERED PRODUCTIVITY TOOLS

Canva introduced Visual Suite 2.0 at Canva Create 2025, integrating AI-driven features like Magic Studio, Canva Sheets, and Magic Charts to streamline design workflows. The suite also includes Canva Code for website creation and an enhanced Photo Editor with AI capabilities. These tools aim to unify design, data, and development in a single platform.

HEYBOSS AI WEBSITE BUILDER (PRODUCT)

HeyBoss offers fully customizable AI websites and apps, unlike Wix and Squarespace's template-based approach.

FULLY OPEN FAST INFERENCE MODELS (HUGGING FACE HUB)

Apriel models from ServiceNow research are designed for fast inference and showcase good performance.

SAMSUNG AND GOOGLE CLOUD EXPAND PARTNERSHIP, BRING GEMINI TO BALLIE, A HOME AI COMPANION ROBOT BY SAMSUNG

Ballie, launching this summer, will provide personalized interactions and proactive home assistance using advanced multimodal reasoning.

building better AI applications

🧑🏫 Automated testing for LLMOps on DeepLearning.AI [6] [Free course]

CHATGPT NOW USES ALL PAST CONVERSATIONS

ChatGPT can now remember all your past conversations, a step toward more personalized and long-term useful AI systems.

MIRA MURATI'S STARTUP AIMS FOR $2B ROUND

Ex-OpenAI execs have joined Mira Murati's new startup, which is targeting a $2B raise despite having no product yet, banking on its star team and ambitious AI goals.

AMAZON'S HEAVY AI INVESTMENT

Amazon CEO Andy Jassy, in their annual letter, emphasizes that massive AI investment is essential to reinvent customer experiences, with over 1,000 GenAI projects underway and growing infrastructure demands.

OMNICAPTIONER

OmniCaptioner is a unified visual captioning framework capable of generating detailed textual descriptions across diverse visual domains, including natural images, textual visuals, and structured graphics. It enhances visual reasoning with LLMs, improves image generation tasks, and enables efficient supervised fine-tuning with less data.

BROWSECOMP AGENT BENCHMARK

OpenAI has released a new agent based benchmark that tests an agent's ability to locate hard-to-find information via actions in the browser. Its DeepResearch system/model gets 51%, whereas humans get around 80%.