The Pulse — September 23, 2026
ES/EN text ready · audio in preparation
- 01GPT-6 Sol and Luna — OpenAIOpenAI (Sep 22)
- 02Better prompt caching for GPT-6 — OpenAIOpenAI (Sep 22)
- 03Claude Opus 5.5 — AnthropicAnthropic (Sep 22)
REGUSCILABS / CUPER
A Newsroom to understand before acting.
CUPER investigates sources and organizes evidence. RegusciLabs turns that context into static, versioned stories that help teams make decisions.
We do not publish noise.
We publish judgment.
Every signal keeps a link to its original material.
We look for what changed and why it may matter.
The strongest signals become decisions, tests, and POCs.
LATEST SIGNALS
What changed.
And what to do about it.
Editorial readings that group several signals around a useful thesis for companies and teams.
AgentsModelsApplied AI
AI stopped answering. Now it wants to act.
Robotics, search, education, and consumer applications are showing the same signal: value is shifting from the model that answers to the system that executes.
Read analysis · 7 min ↗OpenAIAgentsEnterprise
OpenAI Presence confirms that the model is no longer enough.
OpenAI’s new enterprise proposition is not sold as a smarter chatbot, but as a system for operating agents with policies, evaluations, and human escalation.
Read analysis · 6 min ↗OpenAISecurityAgents
When a benchmark reaches production.
An internal cyber-capability evaluation chained vulnerabilities until it reached Hugging Face infrastructure. The lesson is not science fiction; it is systems design.
Read analysis · 7 min ↗ModelsOpen weightsAgents
Inkling and Kimi K3 change the question around open models.
The race is no longer explained by benchmarks alone: long context, multimodality, customization, and sustained real work are beginning to define the choice.
Read analysis · 6 min ↗AnthropicEducationHardware
Claude shows how AI is learning a trade.
One proposal for teachers and another for physical engineering tell the same story: AI becomes valuable when it understands a role’s context, tools, and rules.
Read analysis · 6 min ↗MethodTraining
Training is not the finish line.
Learning can work as a radar: while a team adopts AI, it also reveals which processes are worth transforming.
Read analysis · 5 min ↗HardwareApplied AI
When AI leaves the screen.
Software, sensors, kiosks, 3D printing, and digital fabrication can all be parts of one answer.
Read analysis · 4 min ↗LIVE ARCHIVE
Everything that entered
the radar.
One edition per day.
The complete CUPER-NEWSROOM archive as static, shareable pages with every source visible.
EN
AgentsModelsOpenAI
GPT-6 Sol and Luna — OpenAI
- 02Better prompt caching for GPT-6 — OpenAI
- 03Claude Opus 5.5 — Anthropic
SOURCES
OpenAI (Sep 22) · Anthropic (Sep 22) · arXiv, submitted Sep 22 · Superdesign / open-source repository
EN
ModelsAgentsOpenAI
Gemini 3.8 Live & Live Extended Thinking — Google
- 02ChatGPT Sponsored Agents — OpenAI
- 03Siri AI ships as an OS-level, multimodal agent — Apple
SOURCES
Google Models & Research · OpenAI: Reimagining advertising with AI · Apple Newsroom · xAI: Memory in Grok Build
AUDIOEN
AgentsModelsAnthropic
Claude Projects: a coordinator managing parallel cloud coding sessions
- 02Claude merges chat, Cowork, Docs, Slides, and Design into one surface
- 03Gemini 3.8 Live + Extended Thinking: voice agents that reason while speaking
SOURCES
Anthropic (primary) · Google (primary) · AX / AgentExecutor (primary) · xAI (primary)
AUDIOEN
AgentsAnthropicOpenAI
We Must Pace the Frontier — Dario Amodei / Anthropic
- 02Claude Cowork and chat are now one Claude — Anthropic
- 03Gemini 3.8 Live and 3.8 Live Extended Thinking — Google
SOURCES
Sources linked inside the edition
EN
AgentsOpenAIModels
Gemini 3.8 Live + Extended Thinking: real-time voice agents that can reason, see, and use tools
- 02Claude merges chat, Cowork, Docs, Slides, and Design into one agent workspace
- 03OpenAI starts testing Sponsored Agents inside ChatGPT ads
SOURCES
Google — Sep 15 (updated Sep 17) · Anthropic / Claude — Sep 16 · OpenAI — Sep 16 · Dario Amodei — September 2026
EN
ModelsAgentsOpenAI
Ternary Bonsai 2 27B — near-full 27B capability in 5.9 GB
- 02Bonsai 2’s independent creator test: single-GPU coding, Blender, browser workflows
- 03OpenJev — a local decision model runs in the browser
SOURCES
PrismML (primary) · Bijan Bowen (creator-watch) → PrismML primary above · OpenJev (primary; surfaced on Hacker News) · GitHub project (primary)
AUDIOEN
ModelsAgentsHardware
OpenAI publishes a framework for reporting model misalignment
- 02HarnessTax: how much does the harness matter for coding agents?
- 03NVIDIA introduces native CUDA Rust for GPU kernels
SOURCES
OpenAI (primary) · HarnessTax project / surfaced on Hacker News · NVIDIA Developer Blog (primary) · arXiv paper (primary research)
EN
ModelsAgentsOpenAI
OpenAI Agents API
- 02Muse: Meta’s personal AI agent
- 03Gemini 3.8 Live + Live Extended Thinking
SOURCES
OpenAI — Sept 10 · Meta — Sept 2026 · Google — Sept 15 · Mistral — Sept 16
AUDIOEN
AgentsOpenAIModels
OpenAI — Introducing the Agents API (public beta)
- 02DeepSeek — V4.1-Flash: 552B MoE, 8B/16B active, and cache economics
- 03OpenAI — GPT-Live-1 brings full-duplex voice agents to the API
SOURCES
OpenAI (primary) · https://openai.com/index/introducing-the-agents-api/ · DeepSeek (primary) · https://api-docs.deepseek.com/news/news260910/ · OpenAI (primary) · https://openai.com/index/introducing-gpt-live-1-in-the-api/ · Andon Labs (primary) · https://andonlabs.com/blog/why-we-built-pion
AUDIOEN
AgentsOpenAIModels
OpenAI: Agents API (public beta)
- 02DeepSeek-V4.1-Flash: 552B MoE, 8B/16B active parameters
- 03GPT-Live-1: full-duplex voice agents arrive as an API primitive
SOURCES
OpenAI · Announcement · Docs · DeepSeek · Release notes · Model · Technical report · OpenAI · Announcement · Anthropic · Announcement
AUDIOEN
AgentsOpenAIModels
DeepSeek-V4.1-Flash: a 552B MoE with 8B/16B active parameters
- 02Meta launches Muse, a personal agent with its own secure VM
- 03ChatGPT Images 2.5: faster iterative editing becomes the product
SOURCES
DeepSeek (primary) · https://api-docs.deepseek.com/news/news260910/ · Meta (primary) · https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/ · OpenAI (primary) · https://openai.com/index/introducing-chatgpt-images-2-5/ · OpenAI (primary) · https://openai.com/index/navier-stokes-solution/
AUDIOEN
OpenAIModelsAgents
ChatGPT Images 2.5: faster, more controllable image editing
- 02Meta Muse is a personal agent architecture story, not just an app launch
- 03DeepSeek-V4.1-Flash: an architecture-first cost story
SOURCES
OpenAI — https://openai.com/index/introducing-chatgpt-images-2-5/ · Meta — https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/ · DeepSeek — https://api-docs.deepseek.com/news/news260910/ · Hugging Face / Agnes AI — https://huggingface.co/Agnes-AI/Agnes-3.0-Flash
EMAIL AND TELEGRAM ACTIVE · APP ON THE WAY
THE AI PULSE
What matters today.
In your headphones.
News, context, and RegusciLabs’ practical reading. Get it by email or directly on Telegram; the app will be the next channel.
Choose the channel that works best for you. On Telegram, you can change preferences or unsubscribe directly from the bot.
R
Regu · RegusciLabsThe AI Pulse
TODAY · 07:30
GOOD MORNING, SANTIAGOThree signals to understand what changed while you slept.
- 01OpenAI turns agents into operations.
- 02Anthropic takes Claude into the physical world.
- 03Open models change scale.
AUDIO BRIEFINGThe day in AI
PREVIEW
Information arrives ready to listen to, save, or turn into a decision.