AI pulse last 7 days
Daily AI pulse from YouTube, blogs, Reddit, HN. Ruthlessly filtered.
Sources (41)▶
- criticalAndrej Karpathy
Były dyrektor AI w Tesli, OpenAI cofounder. Każde video to gold.
- criticalAnthropic
Oficjalny kanał Anthropic. Każdy release Claude'a.
- criticalComfyUI Blog
Release log dla integracji ComfyUI — Luma Uni-1, GPT Image 2, ACE-Step music gen, Seedance. Pokrywa video+image+music+workflow.
- criticalOpenAI Blog
Oficjalny blog OpenAI. Wszystkie release.
- criticalSimon Willison's Weblog
Najlepszy 'thinker' AI. Codzienne posty, deep insights, niska hype rate.
- highAI Explained
Głęboka analiza papers i benchmarków, niska hype rate.
- highAI Jason
Praktyczne tutoriale Claude Code, MCP, workflow vibe codingu.
- highBen's Bites
Daily AI digest, creator-friendly tone. Codex, model releases, agentic AI.
- highCole Medin
Vibe coding + agentic workflows + Claude Code MCP integrations.
- highFal AI Blog
Fal hostuje większość nowych AI image/video modeli — ich blog to wczesne sygnały premier.
- highHN: 3D & Gaussian Splatting
HN signal dla 3D generative — Gaussian Splatting, NeRF, image-to-3D. Próg 20 bo niszowa kategoria (top historic 182pts).
- highHN: AI agents / MCP
HN posty o agentach, MCP, vibe codingu z min 100 pkt.
- highHN: Claude / Anthropic
HN posty z 'Claude' lub 'Anthropic' z min 100 pkt.
- highHugging Face Blog
Releases dla image, video, audio, 3D modeli. Część tech-heavy — Gemini relevance odfiltruje noise. Downgraded z critical: za duży volume na 'must-read' status.
- highIndyDevDan
Claude Code power user, prompty, hooki.
- highInterconnects (Nathan Lambert)
AI policy + research analysis. Niska hype rate, opinionated.
- highLatent Space
Podcast + blog Swyx — wywiady z founderami i deep dives engineeringowe.
- highMatt Wolfe
Comprehensive AI tools weekly digest. ~700K subs.
- highMatthew Berman
AI news, model release reviews, agent demos. Wysoki output.
- highr/aivideo
Community AI video — Sora, Veo, Runway, Kling, LTX. Co naprawdę zaskakuje twórców.
- highr/ClaudeAI
Społeczność Claude'a — power users, tipy, problemy.
- highr/LocalLLaMA
Open-source LLMs, lokalne uruchamianie, benchmarks bez hype.
- highr/StableDiffusion
Największa community open-source image gen (700k+ users). Premiery modeli, LoRA, ComfyUI workflows.
- highRiley Brown
Vibe coding, AI builder workflows, Cursor + Claude tutorials.
- highThe Decoder
Niemiecki AI news outlet po angielsku, dobre breaking news.
- highTheo - t3.gg
TypeScript + AI dev workflows. Hot takes, narrative-driven.
- highYannic Kilcher
Paper reviews i deep dives w research AI.
- lowAI Weirdness
Janelle Shane — playful AI experiments, image gen quirks. Niski volume, unikalna perspektywa.
- mediumbycloud
AI papers digestible — między 2MP a Yannic Kilcher.
- mediumCreative Bloq
Design industry — gdzie AI ingeruje w klasyczne dyscypliny graficzne.
- mediumFireship
100-sec format, often AI/LLM + tech news.
- mediumfxguide
VFX i film industry — coraz więcej AI w pipeline. Profesjonalna perspektywa.
- mediumGreg Isenberg
Solo founder vibe — buduje produkty z AI, podcasty z indie hackers.
- mediumr/ChatGPTCoding
Vibe coding tipy, IDE setupy, prompty. Mix wszystkich modeli.
- mediumr/comfyui
ComfyUI workflows — custom nodes, JSON workflows, optymalizacje.
- mediumr/midjourney
Midjourney community — premiery v7+, style references, prompt patterns.
- mediumr/runwayml
Runway-specific community — premiery features, prompt patterns, comparisons z konkurencją.
- mediumr/SunoAI
Suno music gen community — nowe wersje modelu, lyric prompting techniques. Audio AI ma slaby RSS ecosystem.
- mediumTina Huang
AI workflows for data science, practical applications.
- mediumTwo Minute Papers
Krótkie streszczenia papers AI, świetne dla szybkiego scan'a.
- mediumWes Roth
AI news z bardziej clickbaitowym tonem — filtr Gemini odsiewa hype.
Gemini Omni, Gemini 3.2 Flash, a 12M Context Window Model, Claude Replaces Analysts, & More! AI NEWS
A massive week of AI updates including a 12M context window model, GPT-5.5 Instant, and Claude's automation of financial analyst roles.
This week saw a flurry of AI announcements ahead of Google IO, headlined by the leak of Gemini 3.2 Flash and a new Omni model for native video generation. A startup called SubQ introduced a sub-quadratic sparse attention architecture, enabling a staggering 12-million-token context window with 52x faster processing than traditional methods. OpenAI quietly rolled out GPT-5.5 Instant, a faster, more reliable version of their flagship model optimized for real-time use. Anthropic launched specialized Claude agent templates designed to automate entry-level financial analyst tasks, including valuation and market research. Additionally, Google updated Gemma 4 with multi-token prediction for 3x speed gains and enhanced Notebook LM with advanced mind-mapping features.
AI Jason·news·05/06/2026, 06:30 AM·WorldofAI▶Watch here
[AINews] Silicon Valley gets Serious about Services
OpenAI released GPT-5.5 Instant as the new ChatGPT default, while major AI labs are pivoting to professional services to help enterprises deploy agents at scale.
OpenAI has launched GPT-5.5 Instant, a major upgrade in factuality and image understanding, now the default for ChatGPT. Simultaneously, OpenAI and Anthropic are launching multi-billion dollar service joint ventures with private equity firms like Blackstone and Bain Capital. These entities will handle the "last mile" of enterprise AI deployment, focusing on systems integration and workflow modernization. For developers, OpenAI released an Agents SDK for TypeScript, and Cursor introduced automated CI failure fixing. Meanwhile, Meta's new ProgramBench reveals that AI still struggles to generate entire software repositories from scratch, scoring 0% on perfect end-to-end generation of complex projects.
Latent Space·news·05/06/2026, 05:40 AMdatasette-referrer-policy 0.1
A new Datasette plugin to control Referrer-Policy headers, created with AI to fix OpenStreetMap tile loading issues.
Simon Willison released datasette-referrer-policy 0.1, a plugin designed to customize HTTP Referrer-Policy headers in Datasette instances. The project was prompted by a bug in a global power plants demo where OpenStreetMap tiles failed to load because OSM blocks requests from sites using the "no-referrer" policy, which is Datasette's default setting. To solve this without changing global defaults for all users, Willison used Codex and GPT-5.5 to generate the plugin code. This release highlights the utility of AI in quickly spinning up small, single-purpose architectural fixes for web applications and managing specific HTTP security headers.
Simon Willison's Weblog·tooling·05/05/2026, 11:44 PM
ChatGPT update rolls out GPT-5.5 Instant with fewer hallucinations and more personalized answers
GPT-5.5 Instant is now the ChatGPT default, offering a 50% reduction in hallucinations and new 'memory sources' for better transparency.
OpenAI has replaced the default ChatGPT model with GPT-5.5 Instant, emphasizing reliability and transparency. Internal benchmarks report a 52.5% reduction in hallucinations for high-risk topics like medicine and law. A new feature called 'memory sources' has been introduced, allowing users to see exactly which stored context or past interactions influenced a specific response. While the model is rolling out to all users immediately, advanced personalization features involving Gmail and file integration are currently exclusive to Plus and Pro subscribers. This update marks a significant step in grounding LLM outputs in verifiable user data.
The Decoder·model_release·05/05/2026, 06:04 PM·Matthias Bastian
GPT-5.5 Instant System Card
OpenAI releases technical documentation and safety benchmarks for GPT-5.5 Instant, highlighting improved reasoning and low-latency performance.
OpenAI has published the System Card for GPT-5.5 Instant, a new iteration in their model lineup focused on speed and efficiency. The document details the safety evaluations, red-teaming efforts, and performance benchmarks conducted before the public rollout. Key highlights include significant improvements in reasoning capabilities compared to previous small-scale models. The report also addresses potential risks such as bias and misuse, outlining the mitigation strategies implemented. This release marks a shift towards more transparent reporting on model behavior and safety guardrails for high-speed LLM applications.
OpenAI Blog·model_release·05/05/2026, 10:00 AM
GPT-5.5 Instant: smarter, clearer, and more personalized
OpenAI updates ChatGPT with GPT-5.5 Instant, offering higher accuracy, fewer hallucinations, and better personalization for daily use.
OpenAI has officially released GPT-5.5 Instant, positioning it as the new standard model for ChatGPT. The update emphasizes improved reasoning capabilities and a notable decrease in hallucination rates for factual queries. A key addition includes more granular personalization controls, giving users better agency over the model's tone and memory. This release bridges the gap between major version leaps, focusing on speed and reliability for high-frequency tasks. It represents OpenAI's continued push to make LLM interactions feel more natural and context-aware.
OpenAI Blog·model_release·05/05/2026, 10:00 AM
GPT-5.5 VERIFIED Opus 4.7: A Pi Coding Agent That REVIEWS Like YOU
Automate your output reviews by using a specialized 'Verifier' agent that checks the 'Builder' agent's work against strict rules and autonomously triggers fixes.
The video introduces a 'Verifier Agent' architecture designed to overcome the 'review constraint' in agentic workflows. Using a setup with Claude Opus 4.7 as a builder and GPT 5.5 as a verifier, the author demonstrates a system where the second agent automatically audits the first's output via a Unix socket. The verifier checks both claims (e.g., file existence) and visual constraints (e.g., text density in generated diagrams) using advanced vision capabilities. If a rule is violated, the verifier autonomously re-prompts the builder to fix the error without human intervention. This 'one agent, one purpose' approach aims to trade tokens for human time, creating a positive feedback loop for engineering standards.
IndyDevDan·tutorial·05/04/2026, 01:00 PM·IndyDevDan▶Watch here
US wants Claude all to itself... because it's "TOO DANGEROUS"
The US government is treating frontier models like Claude Mythos and GPT 5.5 as national security assets, restricting access due to their autonomous cyber-attack capabilities.
The White House has reportedly intervened to block Anthropic from expanding access to its Claude Mythos model, citing national security risks and compute priority. This follows findings from the UK AI Security Institute (AISI) showing Mythos and OpenAI’s GPT 5.5 can complete complex, multi-step cyber attacks end-to-end. In one test, a task taking a human expert 20 hours was completed in 10 minutes for less than $2 in API costs. This marks a shift from AI as a service to AI as controlled national infrastructure, similar to weapons-grade materials. While experts argue these vulnerabilities were already findable by humans, the concern is the democratization of these skills to non-technical actors globally.
Wes Roth·news·05/01/2026, 04:12 AM·Wes Roth▶Watch here
Relevance auto-scored by LLM (0–10). List shows top 30 from the last 7 days.