#models

• LFM2.5-2.6B: Local agents with 128K context, no cloud required

  LiquidAI's LFM2.5-2.6B runs agentic workflows on-device with 128K context, tool calling, and an OpenAI-compatible endpoint. No cloud inference bill.

  news · models · datasets · open-source

  Aug 4 · 2 min

• 🤗 Kernels: Signing, Trusted Publishers, and Hub Integration

  Hugging Face revamped Kernels with cryptographic signing, trusted publisher verification, and first-class Hub integration. Here's what changed.

  news · models · datasets · open-source

  Jul 6 · 2 min

• Photoroom PRX: Inside Their Training Data Strategy

  Photoroom's PRX Part 4 details their training data strategy — sourcing, curation, synthetic data, and open dataset releases on HuggingFace.

  news · models · datasets · open-source

  Jul 6 · 2 min

• Hugging Face + Cerebras: Gemma 4 now runs speech-to-speech in real time

  Hugging Face and Cerebras built a real-time voice AI pipeline using Google's Gemma 4 model, cutting latency to conversational speeds with open-source components.

  news · models · datasets · open-source

  Jul 1 · 2 min

• OpenAI on AWS: GPT-5.5 and Codex Hit Amazon Bedrock

  OpenAI's frontier models including GPT-5.5 and Codex are now generally available on Amazon Bedrock, letting enterprises use existing AWS procurement and security workflows.

  news · ai-coding · models · Product

  Jun 6 · 2 min

• Codex on Windows finally gets a real sandbox

  OpenAI shipped a native Windows sandbox for Codex, ending the choice between running the agent unconfined or losing it to a Linux VM.

  news · ai-coding · models · Engineering

  May 15 · 1 min

• Veo 3.1 Lite: Google's cheapest video generation model is now in the Gemini API

  Google's Veo 3.1 Lite is available in paid preview via the Gemini API and Google AI Studio. Here's what developers need to know.

  news · models · Developer tools · AI

  May 10 · 2 min

• ChatGPT Workspace Agents: Codex-powered automation for teams

  OpenAI launched workspace agents in ChatGPT for Business, Enterprise, Edu, and Teachers plans. Here's what changed and who should care.

  news · ai-coding · models · product

  May 9 · 2 min

• Google TPU 8i and 8t: Two new chips built for AI agents

  Google announced TPU 8i for fast agentic inference and TPU 8t for large-scale training. Here's what each chip does and who it affects.

  news · models · Google Cloud · AI

  May 9 · 2 min

• OpenAI Academy: Workspace Agents Bring Repeatable Workflows to ChatGPT

  OpenAI Academy published a guide on building workspace agents in ChatGPT for repeatable tasks like briefings, triage, and pipeline monitoring.

  news · ai-coding · models · OpenAI Academy

  May 9 · 2 min

• OpenAI Codex: 4M weekly active users and a push into enterprise

  OpenAI launches Codex Labs and partners with Accenture, PwC, and Infosys to scale Codex across enterprise software development.

  news · ai-coding · models · company

  May 9 · 2 min

• OpenAI Privacy Filter: open-weight PII detection model, runs locally

  OpenAI released Privacy Filter, an open-weight 1.5B parameter model for detecting and redacting PII in text. It runs locally, supports 128K context, and hits 97.43% F1.

  news · openai · privacy · pii

  May 9 · 2 min

• GPT-5.5 Bio Bug Bounty: OpenAI Pays $25K to Break Its Biosafety Guardrails

  OpenAI is paying $25,000 to anyone who can find a universal jailbreak defeating GPT-5.5's five-question bio safety challenge. Applications open now.

  news · ai-coding · models · safety

  May 8 · 2 min

• Microsoft and OpenAI: Amended Deal Ends Revenue Share, Opens Multi-Cloud

  Microsoft and OpenAI have restructured their partnership. Key changes: non-exclusive IP license, no more Microsoft revenue share, and OpenAI can now sell on any cloud.

  news · ai-coding · models · company

  May 8 · 2 min

• OpenAI's Cybersecurity Plan: Five Pillars to Democratize AI Defense

  OpenAI published a five-pillar cybersecurity action plan focused on democratizing AI-powered defense tools for government and commercial defenders.

  news · ai-coding · models · Global Affairs

  May 8 · 2 min

• OpenAI: ChatGPT Enterprise and API Platform clear FedRAMP Moderate

  OpenAI's ChatGPT Enterprise and API Platform have achieved FedRAMP 20x Moderate authorization, opening access to GPT-5.5 for U.S. federal agencies.

  news · ai-coding · models · government

  May 8 · 2 min

• Codex Update: OpenAI Adds Computer Use, Browsing, Image Gen, and Plugins to Its Coding Agent

  OpenAI's Codex coding agent now includes computer use, in-app browsing, image generation, persistent memory, and a plugin system. Here's what changed and what to test.

  news · ai-coding · models · openai

  May 6 · 4 min

• Gemini API: Webhooks Replace Polling for Long-Running Agentic Jobs

  Google adds event-driven webhooks to the Gemini API, eliminating polling overhead for long-running agentic workflows and batch processing tasks.

  news · models · developer-tools · gemini

  May 5 · 2 min

• OpenAI Voice AI: How They Rebuilt WebRTC for Real-Time Conversations at Scale

  OpenAI published a technical deep dive on rebuilding its WebRTC stack to deliver low-latency voice AI globally, powering ChatGPT voice mode's conversational turn-taking.

  news · ai-coding · models · Engineering

  May 5 · 2 min

• OpenAI Stargate: Scaling to Tens of Gigawatts for AGI Compute

  OpenAI is scaling its Stargate data center infrastructure to tens of gigawatts, rethinking chip supply chains and rack design to meet AGI compute demands.

  news · ai-coding · models · infrastructure

  May 5 · 2 min

• Claude Opus 4.7: What Anthropic's Announcement Means for Coding and Agentic Workflows

  Anthropic announced Claude Opus 4.7 targeting coding, agentic tasks, and multi-step reasoning. Here's what developers should evaluate before switching.

  news · models · agents · claude

  May 3 · 5 min

• How to Read OpenAI's GPT-5.5 System Card: What Developers Should Extract and Why

  A practical guide to reading OpenAI's GPT-5.5 system card — which sections matter for developers, what to look for, and how to act on the findings.

  news · ai-coding · models · safety

  May 1 · 6 min

• Codex Automations: How to Build Scheduled and Trigger-Based Workflows with OpenAI's Coding Agent

  Configure OpenAI's Codex agent to run coding tasks on a schedule or in response to events. Step-by-step setup, limits, and alternatives.

  explainer · ai-coding · models · OpenAI Academy

  May 1 · 7 min

• OpenAI Codex: How to Set Up Projects, Threads, and Your First Coding Tasks

  Step-by-step guide to setting up OpenAI Codex projects, threads, and submitting your first coding tasks through the cloud-based agent interface.

  explainer · ai-coding · models · OpenAI Academy

  May 1 · 6 min

• OpenAI on AWS: Codex and Managed Agents now available in enterprise cloud environments

  OpenAI GPT models, Codex, and Managed Agents are now accessible on AWS, letting enterprises build AI coding workflows within their existing cloud infrastructure.

  news · ai-coding · models · openai

  May 1 · 3 min

• OpenAI Codex settings: how to configure personalization, permissions, and detail levels

  A practical guide to configuring OpenAI Codex settings for AI coding workflows, covering personalization, detail levels, and permissions.

  explainer · ai-coding · models · OpenAI Academy

  May 1 · 6 min

• OpenAI Codex: 10 practical use cases for automating work tasks

  OpenAI Academy outlines 10 ways to use Codex at work, from file manipulation to cross-tool integrations. Here is what each use case means in practice.

  explainer · ai-coding · models · OpenAI Academy

  May 1 · 5 min

• GPT-5.5: What OpenAI's Latest Model Means for Coding and Developer Workflows

  OpenAI announced GPT-5.5, targeting multi-step coding, research, and data analysis. Here's what developers need to verify before migrating production workloads.

  news · ai-coding · models · openai

  May 1 · 5 min

• Symphony: OpenAI's open-source spec for turning issue trackers into Codex agent systems

  OpenAI released Symphony, an open-source orchestration spec that connects issue trackers to always-on Codex coding agents, reducing context switching for engineering teams.

  news · ai-coding · models · engineering

  May 1 · 3 min

• OpenAI Responses API: How WebSockets and connection caching cut latency in agentic workflows

  OpenAI details how WebSockets and connection-scoped caching reduced overhead in the Codex agent loop via the Responses API, with practical implications for AI agent developers.

  news · ai-coding · models · engineering

  May 1 · 4 min

• OpenAI Codex: what it is, how it works, and how developers can use it

  OpenAI's new Codex is an agentic task-automation layer, not the old code-generation model. Here's what it does, how to use it, and where it fits.

  explainer · ai-coding · models · OpenAI

  May 1 · 7 min