15,000 AI agents are now waiting to be approved for a credit line

Soon we’ll be getting outbid on houses by an AI agent named LeBotJames

@aipost 🏴

Molts are now discussing crypto markets like they are some regular traders.

@aipost 🏴

So far, this is what we’ve seen from AI agents:

-Running their own social networks
-Talking to each other without humans
-Forming communities and narratives
-Applying for jobs and earning money
-Managing credit and payments
-Living inside simulated worlds with rules and consequences
-Competing in hackathons and building products
-Managing wallets and interacting on-chain
-Creating their own cults and religions
-Talking about creating new languages

Are you entertained?

@aipost 🏴

⚡️Frontier: OpenAI’s platform for deploying AI coworkers in the enterprise

OpenAI has unveiled Frontier, a new platform designed to help companies build, deploy, and manage AI agents that do real work not demos.

What Frontier enables
• Work understanding: Agents learn workflows, handoffs, and context not just prompts
• Tool + computer use: Operate across apps, browsers, and internal systems
• Continuous improvement: Quality increases over time through feedback and iteration
• Governance & observability: Built-in controls, monitoring, and auditability for enterprises

Why it matters
• Moves agents from “chat assistants” to durable digital coworkers
• Targets the hardest enterprise gap: trust, control, and reliability at scale
• Signals OpenAI’s deeper push into enterprise operating infrastructure, not just models

Availability
• Live today for a limited set of customers
• Broader rollout planned over the next few months


Frontier isn’t about smarter chat, it’s about making AI accountable enough to sit inside real org charts.

https://openai.com/index/introducing-openai-frontier/

@aipost 🏴

While OpenAI and Anthropic are engaged in a public PR squabble, Google is setting new records: a year-on-year growth of 17% in search revenue alone was achieved, even though many predicted the end of Google Search with the advent of GPT-4.

While two argue, the third rejoices.

@aipost 🏴

Elon Musk is awarding money for the best Grok imagine videos!

@aipost 🏴

This is what they called AI slop

Everything you see, hear here is generated with kling AI 3.0.

@aipost 🏴

⚡️GPT-5.3-Codex: OpenAI’s most capable model for agentic coding and real computer work

OpenAI just introduced GPT-5.3-Codex, a next-gen Codex model built to handle end-to-end software tasks not just write snippets.

What’s new
• Agentic by design: Handles long, multi-step workflows involving planning, tool use, execution, and iteration
• Stronger real-world coding: Improved reasoning across large codebases, better debugging, testing, and refactoring
• Faster execution: ~25% speed boost vs prior Codex models, optimized for sustained work
• Beyond code: Can operate like a junior teammate helping with research, docs, spreadsheets, deployment, and web builds

How it works
• Interactive by default: You can guide, interrupt, or redirect the model while it’s working
• Computer-use native: Designed to operate across IDEs, CLIs, and web-based tools, not just text boxes

Safety & access
• Classified by OpenAI as high-capability for cybersecurity tasks, with added safeguards and monitoring
• Available now in ChatGPT Codex, IDE extensions, and developer tools
• API access planned for a later release

GPT-5.3-Codex signals OpenAI’s bet that the future of coding models isn’t bigger prompts but agents that can actually do the work.

https://openai.com/index/introducing-gpt-5-3-codex/

@aipost 🏴

🔥 Claude Opus 4.6: Anthropic upgrades its smartest model for long-horizon agentic work

Anthropic introduces Claude Opus 4.6, an upgraded Opus-class model designed to plan more carefully, sustain agentic tasks for longer, and operate reliably across massive codebases while catching its own mistakes.

What’s new
• Stronger planning & self-correction: Plans before acting and detects errors mid-task
• Built for long agentic runs: Sustains complex, multi-step workflows without drifting
• Massive context window: First Opus model with 1M token context (beta)
• Codebase-native: More reliable reasoning across large, real-world repositories

State-of-the-art performance
• Agentic coding
• Multi-discipline reasoning
• Knowledge work
• Agentic search

Product upgrades across Claude
• Claude in Excel: Handles longer-running and harder tasks with better performance
– Supports planning before execution
– Enables advanced features like conditional formatting & data validation
– Applies multi-step changes in a single pass
• Claude in PowerPoint: Expanded capabilities for structured creation
• Claude Code: Deeper agentic coding workflows
• API: More power and flexibility for developers

Availability
• Live today on claude.ai, the Claude Developer Platform, and all major cloud providers
• In Cowork, Opus 4.6 can operate autonomously, putting its full agentic skillset to work on your behalf

Claude Opus 4.6 shows Anthropic’s direction clearly: fewer hand-holds, longer horizons, and agents that can plan, execute, and correct themselves at scale.

https://www.anthropic.com/news/claude-opus-4-6

@aipost 🏴

OpenAI’s blog states that GPT-5.3-CODEX was used to support its own creation.

It achieves SOTA score of 57% at SWE Bench Pro and 76% on TerminalBench.

"With GPT‑5.3-Codex, Codex goes from an agent that can write and review code to an agent that can do nearly anything developers and professionals can do on a computer."

@aipost 🏴

🚨 Anthropic had 16 AI agents build a C compiler from scratch. 100k lines, compiles the Linux kernel, $20k, 2 weeks.

To put that in perspective GCC took thousands of engineers over 37 years to build. (Granted from 1987 - however) One researcher and 16 AI agents just built a compiler that passes 99% of GCC's own torture test suite, compiles FFmpeg, Redis, PostgreSQL, QEMU and runs Doom.

They say they "(mostly) walked away." But that "mostly" is doing heavy lifting.

No human wrote code but the researcher constantly redesigned tests, built CI pipelines when agents broke each other's work, and created workarounds when all 16 agents got stuck on the same bug.

The human role didn't disappear. It shifted from writing code to engineering the environment that lets AI write code.

@aipost 🏴

This guy dropped 6 prompt hacks that make ChatGPT feel 10× smarter.

@aipost 🏴

OpenAI recently posted this:

“We worked with Ginkgo to connect GPT-5 to an autonomous lab, so it could propose experiments, run them at scale, learn from the results, and decide what to try next”

GPT-5 is now proposing experiments, executing them in autonomous labs, learning from the results, and iterating.

This is not software anymore.
This is automated scientific progress.

@aipost 🏴

💰 Big tech plans to spend BIG on AI in 2026

Here’s how much big tech plans to spend on CAPEX in 2026

$200 Billion - Amazon $AMZN
$180 Billion - Google $GOOGL
$125 Billion - Meta Platforms $META
$117.5 Billion - Microsoft $MSFT
$20 Billion - Tesla $TSLA
$13 Billion - Apple $AAPL

@aipost 🏴