The archive · 11 labs · 3,050 dispatches

The labs

No more opening a dozen official blogs every morning. First-hand releases from OpenAI, Anthropic, DeepMind and the rest, each with its substance pulled out.

151

OpenAI partners with CodeAI to launch ChatGPT for Teens and AI literacy programs

OpenAI announced a partnership with CodeAI to launch ChatGPT for Teens and a suite of education programs that teach high‑school students AI literacy, critical thinking, and responsible use.

152

OpenAI Pacing Model Development for Cyber-Critical Capabilities

OpenAI has temporarily slowed the scaling of its latest models, including Astra, to implement strengthened monitoring, alignment, and security safeguards following evidence of critical cybersecurity capabilities.

153

Asana Accelerates Legacy Code Migration Using OpenAI Codex

Asana used OpenAI Codex to remove an outdated testing system in two weeks, completing a project estimated to take five years and $6 million.

154

Sentence Transformers v6.0 Multi-Vector Encoder release

Sentence Transformers v6.0 adds a MultiVectorEncoder model type for ColBERT‑style late‑interaction retrieval, enabling token‑level embeddings across text, images, audio, and video with higher retrieval quality at the cost of larger indexes.

155

NVIDIA Implementation of ChatGPT Work

NVIDIA uses ChatGPT Work to automate operational processes and synthesize industry intelligence, reducing manual analysis time and accelerating prototype development.

156

Claude Protein Design and Analytical Chemistry Capabilities

Anthropic demonstrates that Claude models can autonomously design high-affinity protein binders with hit rates exceeding human experts and automate complex analytical chemistry workflows including NMR and LC-MS data processing.

157

Dharma AI GPU Management: Increasing Cluster Utilization via Constraint-Aware Allocation

Dharma AI developed a constraint-aware GPU allocator that increased GPU utilization by up to 33 percentage points and priority-weighted output by up to 105% compared to FIFO scheduling on identical hardware.

158

The Defender’s Window: OpenAI's Strategy for AI-Driven Cybersecurity

OpenAI outlines a comprehensive defensive strategy to counter AI-powered cyberattacks following the OpenAI-Hugging Face incident, emphasizing the urgent need for organizations to automate security programs using frontier intelligence.

159

OpenAI joins PORTS-Pike project

OpenAI is partnering with SB Energy, NVIDIA, and the U.S. Department of Energy to secure 8 gigawatts-IT of capacity at the PORTS-Pike Technology Campus in Ohio to support frontier AI training and product demand.

160

OpenAI New Policy Ideas for the Intelligence Age Grants

OpenAI is providing $1 million in funding and $1 million in API credits to 14 independent global projects focused on economic opportunity and societal resilience in the AI era.

161

vLLM-Omni Distributed Layerwise Offload

vLLM-Omni introduces Distributed Layerwise Offload (DLO), enabling the efficient serving of large Diffusion Transformer (DiT) models over 200B parameters by optimizing HBM and host memory usage through weight sharding and double-buffered prefetching.

162

Claude Text Watermarking Implementation

Anthropic is implementing text watermarking in future Claude models to comply with the EU AI Act, using a method that alters the source of randomness in word selection without affecting output quality.

163

vLLM Adaptive Verification with DSpark

vLLM introduces adaptive verification using DSpark's confidence head to dynamically adjust the number of speculative tokens verified per step, maintaining high throughput across varying concurrency levels.

164

State of Open Models Summer 2026 Report

Hugging Face’s Summer 2026 report shows Chinese labs now dominate frontier open‑model releases, small models still capture most downloads, Qwen has become the community’s base model, and agents have become the primary Hub users.

165

Grok 4.6 in GitHub Copilot

xAI has integrated Grok 4.6, its latest coding model, into GitHub Copilot, making it available for developers using VS Code and GitHub.

166

Strands Robots and LeRobot streaming data loop with Hugging Face Storage Buckets

Hugging Face announced a full data loop that lets a Strands robot record LeRobot demonstrations, sync them to a mutable Hugging Face Storage Bucket with byte‑level deduplication, stream the dataset directly from the Hub for training, and deploy the resulting policy back to hardware—all without local downloads.

167

Gemini 3.7 Flash release notes / what's new

Google DeepMind has released Gemini 3.7 Flash, a model optimized for coding and agents that offers significant performance gains over Gemini 3.6 Flash at half the introductory cost.

168

Fyxer AI Executive Assistant Technical Implementation

Fyxer built a highly contextual AI executive assistant using a system of 30-50 specialized OpenAI models and a dataset of 500,000 hours of human assistant workflows to achieve a 90% 90-day user retention rate.

169

GPT-5.6 Release Notes: Advancing Agent Price-Performance

OpenAI has released the GPT-5.6 model family, which significantly reduces the cost of frontier-level agent performance through improved model selection, new API controls for reasoning continuity, and programmatic tool calling.

170

OpenAI GPT-5.6 Sol Ultrafast Mode Preview

OpenAI has introduced Ultrafast mode for GPT-5.6 Sol, a new service tier powered by Cerebras that delivers up to 750 output tokens per second, increasing processing speed by up to 14x over standard processing.

171

OpenAI Appoints Dali Rajic as Chief Revenue Officer

OpenAI has appointed Dali Rajic as Chief Revenue Officer to scale its global revenue organization as the company reaches over one billion weekly active users.

172

Hugging Face ICML 2026 Open Reproductions Report

Hugging Face coordinated a community hackathon using coding agents to reproduce 2,226 ICML 2026 papers, finding that 51% had at least one verified claim while 23% had at least one falsified or contested claim.

173

DeepSeek-V4-Pro GA Release

DeepSeek has released DeepSeek-V4-Pro, featuring enhanced agent capabilities, flexible reasoning effort settings, and native OpenAI Responses API support.

174

Anthropic Patterns and Problems in Multiagent Systems

Anthropic research reveals that while frontier models can coordinate on parallel tasks, they struggle with peer-level collaboration, exhibit systemic failures due to behavioral conformity, and can escalate into 'turf wars' when given incompatible goals.

175

OlmoEarth Embeddings: Custom Vector Exports for Earth Observation

Hugging Face and AllenAI have introduced custom embedding exports in OlmoEarth Studio, allowing users to generate compact numerical representations of Earth-observation data for downstream geospatial analysis.

176

Google DeepMind SL2T: Sign-Language-to-Text Translation for Pixel 11

Google DeepMind has introduced SL2T, a multilingual sign-language-to-text model that enables sign-to-text dictation in Gboard and Live Transcribe on Pixel 11, starting with American Sign Language (ASL).

177

LFM2.5-VL-3B release notes / what's new

Liquid AI has released LFM2.5-VL-3B, a 3.1B parameter vision-language model optimized for edge devices, featuring improved screen understanding, grounding, and tool calling.

178

OpenAI Enterprise AI Adoption: From Assistance to Execution

OpenAI reports a widening 'frontier gap' where the top 10% of enterprise users generate 8.3x more output tokens than typical firms by shifting from simple AI assistance to agentic execution.

179

vLLM Day 0 Support for Qwen3.8-2.4T-A95B

vLLM has announced Day-0 support for Qwen3.8-2.4T-A95B, a 2.4-trillion-parameter sparse MoE model that brings Qwen-Max-class capabilities to open-weight releases.

180

RingCentral AI-Native Development and Operational Integration

RingCentral has implemented an AI-native workflow using ChatGPT Work and Codex to accelerate product development and automate PMO operations across both technical and non-technical staff.

181

Anthropic Research: Reviewing the Evidence on Worker Retraining Programs

Anthropic and researcher David Roodman find that while job retraining programs have modest positive effects, they are likely insufficient to mitigate large-scale labor market disruption caused by AI.

182

Grok 4.6 release notes / what's new

xAI has released Grok 4.6, a model optimized for long-running agents, complex coding, and visual work, matching GPT-5.6 Sol on the AA Intelligence Index.

183

ALTK-Evolve vs ACE: Same Lessons, Fewer Tokens

ALTK‑Evolve matches or exceeds ACE’s task‑completion accuracy while using only 20‑40% of the inference tokens, thanks to selective guideline retrieval instead of injecting a full playbook.

184

Mistral AI Sovereign AI Infrastructure and Regional Inference Update

Mistral AI is enhancing AI sovereignty for enterprises and governments by introducing regional inference endpoints, support for third-party open models, and a long-term European compute capacity coalition.

185

OpenAI Daybreak models now available on AWS Bedrock

OpenAI announced that its Daybreak cybersecurity models, including Daybreak Blue (general‑purpose frontier models like GPT‑5.6 Sol) and Daybreak Red (purpose‑trained security models), are now accessible through Amazon Bedrock, enabling enterprises to integrate advanced AI‑driven security capabilities within existing AWS workflows.

186

NVIDIA Nemotron 3.5 Lightning Release

NVIDIA Nemotron 3.5 Lightning is a 30B parameter open model with 3B active parameters per token, designed for local agentic workflows and multi-step tasks with a 1M token context window.

187

xAI Grok Bot Release

xAI has launched Grok Bot, AI teammates capable of operating their own cloud computers to execute end-to-end tasks across various applications and tools without requiring APIs.

188

Anthropic Building Effective AI Agents – Practical Patterns and Guidance

Anthropic outlines practical patterns for building LLM‑augmented agents and workflows, emphasizing simple composable designs, when to use each pattern, and best practices for tool engineering.

189

Anthropic Claude Sonnet 5 announcement

Anthropic announced Claude Sonnet 5, a more agentic, cost‑effective model that rivals Opus‑class performance while improving safety and pricing.

190

OpenAI AI-Native Finance Function Implementation

OpenAI is redesigning its finance operations to achieve a zero-day close and continuous forecasting by integrating AI into core workflows and empowering finance professionals to build their own tools.

191

NVIDIA Magpie TTS Multilingual Release

NVIDIA has released Magpie TTS Multilingual, a 364M-parameter open-weights model supporting 12 languages designed for low-latency, production-ready voice agents.

192

OpenAI's Proposal for Responsible AI Infrastructure in Texas

OpenAI has sent a letter to Texas Governor Greg Abbott outlining its commitment to developing responsible AI infrastructure within the state to ensure meaningful benefits for Texans.

193

Model ML and GPT-5.6 Sol for Finance Workflow Automation

Model ML utilizes GPT-5.6 Sol to automate the end-to-end finance workflow, significantly improving the professional-readiness rate of PowerPoint and Excel deliverables compared to previous models.

194

Efficient Knowledge Distillation for LLMs: Offline Top-K Logits and Fused Chunked KL Loss

Multiverse Computing introduces a memory-efficient distillation method using offline top-K logit caching and a fused chunked KL loss to significantly reduce VRAM requirements and training costs for Large Language Models.

195

OpenAI Daybreak Cyber Partner Program

OpenAI has launched the Daybreak Cyber Partner Program to provide security partners with access to frontier cyber models to help organizations find and fix vulnerabilities more efficiently.

196

OpenAI Daybreak and GPT-5.6-Cyber Release

OpenAI has expanded the Daybreak program and introduced GPT-5.6-Cyber, a specialized model designed to provide trusted defenders with advanced cybersecurity capabilities and reduced refusals for authorized security research.

197

Meta Muse Glimmer 30B Release

Meta has released Muse Glimmer, a 30B parameter multimodal model distilled from Muse and released under Apache 2.0, optimized for local agentic use cases like coding and document analysis.

198

ChatGPT Business Premium Seats Release

OpenAI has introduced Premium seats for ChatGPT Business, offering 5x more usage and the removal of the five-hour usage limit for high-capacity users.

199

Virgin Atlantic Integration of ChatGPT Work

Virgin Atlantic is utilizing ChatGPT Work to accelerate competitive research, unify customer journey data, and streamline product planning to improve the end-to-end passenger experience.

200

Zapier Marketing Automation with ChatGPT Work

Zapier's enterprise marketing team uses ChatGPT Work to automate lead funnel optimization and campaign execution, resulting in seven-figure monthly pipeline growth.