Weekly Digest — 2026 W14
Weekly Digest — 2026 W14 2026-03-30 – 2026-04-05 | 7 entries This Week in Numbers By Signal Signal Count 🟡 notable 4 🟢 incremental 3 By Category Category Count Model Release 3 Infrastructure 2...
awesome-ai-radar
Track what matters. Ignore the noise.
What's Hot Right Now
game-changer
GLM-5: First Frontier Model Trained Entirely on Non-NVIDIA Hardware
notable
Gemini 3.1 Pro Doubles Reasoning with 77.1% ARC-AGI-2
notable
DeerFlow 2.0: ByteDance's Open-Source SuperAgent Harness
notable
Vercel AI SDK 6 Adds Agents, Tool Approval, and Full MCP Support
game-changer
Alibaba's Qwen 3.5 9B Outperforms OpenAI's 120B on Graduate-Level Reasoning
All Entries
Sorted by relevance — most important first
Weekly Digest — 2026 W13
Weekly Digest — 2026 W13 2026-03-23 – 2026-03-29 | 16 entries This Week in Numbers By Signal Signal Count 🟡 notable 9 🟢 incremental 7 By Category Category Count Tool 5 Model Release 4 Framework 2...
Intercom Fin Apex 1.0: Domain-Specific Model Outperforms GPT-5.4 and Opus on Customer Service
backend
AI-agents
enterprise
What happened Intercom shipped Fin Apex 1.0, a custom-trained model that now handles approximately 100% of English chat and email customer conversations on their platform. Built by a 60-person AI team...
Shannon: Autonomous AI Pentester Explodes to 35K GitHub Stars
security
developer-tools
testing
What happened Shannon, an open-source autonomous white-box AI pentester for web applications and APIs, crossed 34,000 GitHub stars in March 2026 after gaining over 21,000 stars in a single month. The...
US States Rush to Regulate AI While Congress Remains Deadlocked
policy
enterprise
compliance
What happened NPR reported on March 28 that US states are increasingly passing their own AI regulations — covering child safety, transparency requirements, and whistleblower protections — while...
Waymo Reaches 500K Weekly Paid Rides Across 10 US Cities
autonomous-vehicles
infrastructure
AI-deployment
What happened Waymo announced it now provides 500,000 paid robotaxi rides per week across 10 US cities, representing tenfold growth from 50,000 weekly rides in May 2024. The fleet of approximately...
Apple iOS 27 Opens Siri to Third-Party AI Chatbots via Extensions
mobile
AI-platform
frontend
What happened Apple plans to allow third-party AI chatbots to integrate with Siri in iOS 27, according to a Bloomberg report on March 26. Users with apps like Claude, Gemini, or other AI assistants...
Meta TRIBE v2: Foundation Model Predicting Human Brain Responses to Sights and Sounds
research
neuroscience
AI-safety
What happened Meta FAIR released TRIBE v2 on March 26, a trimodal brain encoding model that predicts fMRI brain activity in response to video, audio, and text stimuli. Trained on 451.6 hours of fMRI...
Anthropic Claude Mythos Leak Reveals Step-Change Model Above Opus
backend
security
AI-platform
What happened A configuration error in Anthropic's content management system exposed close to 3,000 unpublished assets, including a draft blog post announcing Claude Mythos — a new model tier above...
Gemini 3.1 Flash Live: Real-Time Audio-to-Audio Model for Voice Agents
backend
voice
AI-agents
What happened Google released Gemini 3.1 Flash Live on March 26, a dedicated audio-to-audio model optimized for real-time dialogue and voice-first AI applications. The model offers improved acoustic...
Mistral Voxtral TTS: First Open-Weight Frontier Text-to-Speech Model
backend
voice
open-source
What happened Mistral released Voxtral TTS on March 26, a 4B-parameter open-weight text-to-speech model comprising a 3.4B transformer decoder, a 390M flow-matching acoustic transformer, and a 300M...
Google Lyria 3 Pro: 3-Minute Music Generation with Structural Awareness
creative
multimedia
API
What happened Google released Lyria 3 Pro on March 25, an advanced music generation model from DeepMind that creates tracks up to 3 minutes long, a 6x increase from Lyria 3's 30-second limit. The...
OpenAI Launches Safety Bug Bounty for AI Agent Vulnerabilities
security
AI-agents
MCP
What happened OpenAI launched a public Safety Bug Bounty program on March 25, hosted on Bugcrowd, focused on identifying AI abuse and safety risks that fall outside traditional security...
ProofShot: Visual Verification CLI for AI Coding Agents
developer-tools
testing
AI-agents
What happened ProofShot, an open-source CLI tool that gives AI coding agents "eyes" to verify UI they build, gained traction on Hacker News on March 25. The tool opens a real browser,...
Claude Cowork Gets Computer Use and Cross-Device Dispatch
developer-tools
productivity
desktop
What happened Anthropic launched computer use capabilities for Claude Cowork on March 24, enabling the desktop agent to open apps, navigate browsers, fill spreadsheets, and interact with on-screen...
ChatGPT Launches Library: Persistent File Storage Across Conversations
productivity
frontend
data
What happened On March 23, 2026, OpenAI rolled out the Library feature to ChatGPT for Plus, Pro, and Business users globally (with EEA, Switzerland, and UK availability pending). Files uploaded or...
Dapr Agents v1.0 Reaches General Availability for Production AI Agent Deployments
agents
backend
infrastructure
kubernetes
enterprise
What happened On March 23, 2026, the CNCF announced the general availability of Dapr Agents v1.0 at KubeCon + CloudNativeCon Europe in Amsterdam. The project — a year-long collaboration between...
Weekly Digest — 2026 W12
Weekly Digest — 2026 W12 2026-03-16 – 2026-03-22 | 40 entries This Week in Numbers By Signal Signal Count 🔴 game-changer 3 🟡 notable 24 🟢 incremental 12 ⚪ noise 1 By Category Category Count Tool 14...
GPT-5.2-Codex: OpenAI's Most Advanced Coding Model with Long-Horizon Agentic Reasoning
coding
devtools
agents
backend
security
What happened On March 22, 2026, OpenAI released GPT-5.2-Codex as a specialized coding model built for long-horizon agentic programming tasks. The model achieves state-of-the-art performance on...
Crius: Autonomous Code Improvement Engine with Self-Improving Prompts
devtools
backend
frontend
What happened Crius is a new open-source autonomous code improvement engine that runs six specialized AI agents in a continuous loop on a server with Claude Code. You register a project, set a goal,...
OpenGitClaw: Autonomous GitHub Repository Maintenance Agent
backend
devops
What happened OmegaCore Labs released OpenGitClaw on March 20, 2026 — an autonomous agent that maintains GitHub repositories around the clock. It operates using a smart planner that builds multi-step...
White House Unveils Six-Point National AI Legislative Framework
policy
enterprise
infrastructure
What happened On March 20, the Trump administration released a comprehensive national AI legislative framework addressing six objectives: protecting children and empowering parents, safeguarding...
Adobe Firefly Custom Models Enter Public Beta
design
frontend
What happened Adobe opened its Firefly Custom Models to public beta on March 19, 2026. Creators can now upload 10-30 of their own images to train a custom model that captures their specific...
Cloudflare Workers AI Enters the Frontier Model Game with Kimi K2.5
backend
infrastructure
agents
What happened On March 19, 2026, Cloudflare announced that Workers AI now supports frontier-scale open-source models, starting with Moonshot AI's Kimi K2.5. This is the first time a model of this...
OpenAI Acquires Astral: Ruff, uv, and ty Join the Codex Ecosystem
python
devtools
backend
open-source
What happened On March 19, 2026, OpenAI announced it will acquire Astral, the startup behind the Python developer tooling suite used by millions of developers. Astral's tools — Ruff (the fast Python...
AI Search Engines Are Not Replacing Google (Yet)
enterprise
infrastructure
What happened Multiple outlets ran headlines this week claiming AI search engines (Perplexity, SearchGPT, Gemini Search) are "killing Google Search" based on a SimilarWeb report showing...
Claude Opus 4.6 Ships with 1M Token Context
claude-code
api
agents
enterprise
What happened Anthropic released Claude Opus 4.6 with a native 1 million token context window, making it the first frontier model to ship with megascale context as a default capability rather than an...
Google Stitch Introduces Vibe Design: AI-Native UI Canvas with Voice Interaction
frontend
mobile
design
What happened Google redesigned Stitch from Google Labs on March 18, 2026, transforming it from a text-to-UI generator into an AI-native design platform. The major update introduces an infinite...
VS Code 1.112 Ships MCP Server Sandboxing and Expanded Agent Autonomy
devtools
frontend
backend
What happened Visual Studio Code 1.112 released on March 18 with three significant AI-related features: sandboxed MCP servers on macOS and Linux (restricting file system and network access for locally...
Anthropic Publishes Largest Qualitative AI User Study: 81,000 Participants
product
research
design
What happened On March 18, 2026, Anthropic published results from the largest and most multilingual qualitative study of AI users to date. Nearly 81,000 Claude.ai users participated, sharing how they...
Anthropic Models API Now Returns Capability Fields for Programmatic Discovery
api
backend
devtools
What happened On March 18, 2026, Anthropic updated the Models API (GET /v1/models and GET /v1/models/{model_id}) to return structured capability metadata for each model. Responses now include...
Gemini API Adds Built-in Tool and Function Calling Combination for Gemini 3
backend
agents
What happened On March 18, 2026, Google released a new Gemini API feature that allows developers to use built-in tools (Google Search, Google Maps) alongside custom function calling tools in a single...
Xiaomi MiMo-V2-Pro Revealed as Mystery Model Hunter Alpha After Topping OpenRouter Charts
backend
agents
coding
What happened Xiaomi's MiMo team revealed MiMo-V2-Pro on March 18, 2026 — a 1-trillion parameter Mixture-of-Experts foundation model with 42 billion active parameters per inference. A week earlier, an...
Claude Code Ships 16 Releases in March: Voice Mode, /loop, 128k Output
developer-tools
cli
agents
plugins
What happened Anthropic released 16 versions of Claude Code in March 2026 (v2.1.63 through v2.1.79), introducing push-to-talk voice mode via /voice, recurring task scheduling with /loop, configurable...
Claude Cowork Dispatch: Persistent Agent Threads from Phone to Desktop
backend
frontend
devops
mobile
What happened Anthropic launched Dispatch for Claude Cowork on March 17, 2026 in research preview. Dispatch creates a single persistent conversation thread between the Claude mobile app and Claude...
GitHub Copilot Agent Mode Exits Preview
devtools
agents
enterprise
What happened GitHub Copilot's agent mode graduated from technical preview to general availability for all Copilot Enterprise and Business subscribers. The feature allows Copilot to autonomously plan...
OpenAI Releases GPT-5.4 Mini and Nano for High-Volume Workloads
api
cost-optimization
edge-deployment
What happened OpenAI released GPT-5.4 mini and GPT-5.4 nano on March 17, 2026. Mini runs more than 2x faster than GPT-5 mini and approaches GPT-5.4 on benchmarks including SWE-Bench Pro, priced at...
MCP Server Ecosystem Crosses 10,000 Published Servers
claude-code
agents
devtools
open-source
What happened The Model Context Protocol (MCP) ecosystem has crossed 10,000 published servers on the community registry, up from roughly 1,000 at the start of 2025. Major additions in the past month...
Mistral Launches Forge: Enterprise Custom Model Training Platform
backend
enterprise
mlops
What happened Mistral AI announced Forge at NVIDIA GTC on March 17, a platform that lets enterprises train custom AI models from scratch on their own proprietary data. Unlike competitors that rely on...
LangChain Releases v1.0 After Two Years of Breaking Changes
agents
rag
open-source
devtools
What happened LangChain shipped v1.0 with a stable API guarantee, a significantly simplified core (down from 400+ classes to under 80), and a clear migration path from the 0.x series. The release...
Mistral Small 4 Unifies Instruct, Reasoning, and Multimodal in One Open-Source MoE
open-source
self-hosting
multimodal
agents
What happened Mistral AI released Mistral Small 4 on March 16, 2026 — a 119-billion-parameter Mixture-of-Experts model with 128 experts and a 256k context window, licensed under Apache 2.0. The model...
NVIDIA Launches Agent Toolkit and OpenShell for Enterprise AI Agents
backend
devops
infrastructure
What happened At GTC on March 16, 2026, NVIDIA announced the Agent Toolkit — an open-source collection of building blocks for enterprise AI agents. The toolkit includes OpenShell, an open-source...
obra/superpowers: Agentic Skills Framework Reaches 92k Stars
devtools
backend
frontend
What happened Superpowers, an open-source agentic skills framework by Jesse Vincent, reached 92,000 GitHub stars and topped the GitHub trending page on March 18. The framework provides a complete...
Hindsight: MCP-Native Agent Memory Scoring 91% on LongMemEval
agents
mcp
claude-code
memory
open-source
What happened Vectorize.io's Hindsight hit GitHub Trending after a rapid-fire week of integrations: MCP server support (March 4), Pydantic AI (March 9), and Ollama (March 10). The open-source agent...
OpenAI Cuts GPT-5 Turbo Pricing by 40%
api
enterprise
infrastructure
What happened OpenAI reduced GPT-5 Turbo API pricing by 40%, bringing input tokens to $1.50/M and output tokens to $6/M. The cut follows Google's Gemini 2.5 Pro pricing announcement last month and...
EU Council Agrees to Streamline AI Act with Extended Timelines and SME Relief
policy
enterprise
What happened On March 13, the EU Council agreed its position on streamlining the AI Act as part of the Omnibus VII simplification package. Key changes include extending the timeline for high-risk AI...
Anthropic Invests $100M in Claude Partner Network for Enterprise Adoption
enterprise
backend
What happened Anthropic launched the Claude Partner Network on March 12 with a $100 million investment commitment for 2026. The program provides training, technical support, and joint market...
Anthropic Launches The Anthropic Institute for AI Safety and Societal Impact Research
policy
safety
research
What happened Anthropic launched The Anthropic Institute on March 11, 2026, led by co-founder Jack Clark as Head of Public Benefit. The institute consolidates three existing research teams: the...
Meta Reveals Four MTIA Chip Generations by 2027 for AI Inference at Scale
infrastructure
backend
mlops
What happened Meta published its MTIA chip roadmap on March 11, revealing four new chip generations — MTIA 300, 400, 450, and 500 — to be deployed by end of 2027. MTIA 300 is already in production,...
NVIDIA Nemotron 3 Super: Hybrid Mamba-Transformer MoE for Agentic AI
agents
inference
self-hosting
enterprise
What happened NVIDIA announced Nemotron 3 Super at GTC 2026 — a 120B parameter hybrid model combining Mamba, Transformer, and MoE architectures with only 12B active parameters per token. It scores...
Perplexity Personal Computer: Always-On AI Agent with Local File Access
agents
desktop
enterprise
macos
What happened At its Ask 2026 developer conference, Perplexity unveiled "Personal Computer" — software that runs continuously on a dedicated Mac mini, giving Perplexity's cloud-based AI...
MCP 2026 Roadmap: Horizontal Scaling, Enterprise Auth, and .well-known Discovery
mcp
infrastructure
enterprise
agents
What happened The Model Context Protocol maintainers published the 2026 roadmap outlining four priorities: evolving transport and session models for horizontal scaling without holding state,...
Google Releases Gemini Embedding 2: First Multimodal Embedding Model
backend
search
data
What happened Google released gemini-embedding-2-preview on March 10, 2026 — its first multimodal embedding model. Unlike traditional text-only embedding models, Gemini Embedding 2 supports text,...
Anthropic Launches Multi-Agent Code Review for Claude Code
backend
frontend
devops
What happened Anthropic launched Code Review as a new Claude Code capability on March 9, 2026. When a pull request opens on GitHub, the system dispatches multiple specialized AI agents in parallel —...
Anthropic Sues Pentagon After Refusing Mass Surveillance and Autonomous Weapons Use
policy
enterprise
safety
governance
What happened According to TechCrunch, Anthropic sued the U.S. Department of Defense after the Pentagon labeled the AI company a "supply-chain risk" — a designation that followed Anthropic's...
xAI Launches Grok 4.20 with Built-In Multi-Agent Debate System
agents
reasoning
hallucination-reduction
What happened xAI released the Grok 4.20 Beta on March 9, 2026 in three variants: Non-Reasoning, Reasoning, and Multi-Agent. All three support a 2-million-token context window. The Multi-Agent variant...
VS Code 1.111 Debuts Weekly Releases, Autopilot Mode, and Agent Permissions
ide
developer-tools
agents
copilot
What happened Microsoft released VS Code 1.111 on March 9, 2026, marking the start of a weekly Stable release cadence. The release introduces a three-tier agent permission system: Default Approvals...
Karpathy's autoresearch Hits 23k Stars in 3 Days: AI Agents Running ML Research
ml-research
agents
open-source
training
What happened Andrej Karpathy released autoresearch on March 6, 2026, a tool that enables AI agents to automatically run nanochat (small chat model) training research on a single GPU. As of March 9,...
Lightricks Ships LTX 2.3: Open-Source 4K Video with Synchronized Audio
video-generation
open-source
creative-tools
What happened Lightricks released LTX 2.3, a Diffusion Transformer model that generates synchronized video and audio in a single pass. The model supports resolutions up to 4K at 50 FPS and can produce...
GPT-5.4 Ships in Three Variants with 1.05M Context
api
chatgpt
enterprise
What happened OpenAI released GPT-5.4 in three variants — Standard, Thinking, and Pro — supporting context windows up to 1.05 million tokens. The model reduces individual claim errors by 33% and...
Cursor Adds JetBrains IDE Support
ide
developer-tools
jetbrains
What happened Cursor became available in JetBrains IDEs in early March 2026 through the Agent Client Protocol (ACP). Developers can install the Cursor ACP from the JetBrains ACP Registry and...
Alibaba's Qwen 3.5 9B Outperforms OpenAI's 120B on Graduate-Level Reasoning
open-source
edge-deployment
efficiency
multimodal
What happened Alibaba's Qwen Team released the Qwen 3.5 Small series on March 2, 2026 — four models at 0.8B, 2B, 4B, and 9B parameters, all under Apache 2.0. According to the official model card, the...
Vercel AI SDK 6 Adds Agents, Tool Approval, and Full MCP Support
typescript
nextjs
agents
mcp
frameworks
What happened Vercel released AI SDK 6 with the v3 Language Model Specification that powers native agent support, tool execution approval flows, built-in DevTools, and full MCP (Model Context...
DeerFlow 2.0: ByteDance's Open-Source SuperAgent Harness
agents
open-source
backend
python
What happened ByteDance released DeerFlow 2.0, a ground-up rewrite of their open-source SuperAgent framework. Built on LangGraph and LangChain, it ships with sub-agent orchestration, sandboxed...
Gemini 3.1 Pro Doubles Reasoning with 77.1% ARC-AGI-2
api
agents
enterprise
reasoning
What happened Google released Gemini 3.1 Pro in preview, scoring 77.1% on ARC-AGI-2 — more than double the reasoning performance of Gemini 3 Pro. The model ships with a 1M token context window and is...
GLM-5: First Frontier Model Trained Entirely on Non-NVIDIA Hardware
open-source
self-hosting
api
agents
infrastructure
What happened Zhipu AI (z.ai) released GLM-5, a 744B MoE model with 40B active parameters, under the MIT license. It was trained entirely on 100,000 Huawei Ascend 910B processors — zero NVIDIA...