← Tech / AI / IT Monitor Index Tech / AI Generated 2026-06-11 19:30 UTC

Tech / AI / IT Monitor

June 11, 2026 · Based on tweets from the last 24 hours · 156 tweets analyzed · model: ollama-cloud/glm-5.1:cloud

Daily Intelligence Briefing — Tech / AI / IT Monitor

Date: 2026-06-11 | Reporting Period: Last 24 Hours

Executive Summary

The past 24 hours were dominated by the launch of Anthropic's Fable 5 model, which immediately sparked intense controversy over safety policy changes that appear to deliberately limit the model's AI research capabilities—galvanizing the open-source AI community. Meanwhile, Nous Research's Hermes Agent saw significant feature updates including remote file access, unified profile management, and a new Write Gate for memory/skill approvals. On the applied-AI front, developers demonstrated extraordinary productivity gains: @levelsio used Fable to port Return to Castle Wolfenstein (2001) to web/WASM with multiplayer in ~1 hour, and @jezell burned 15B+ tokens porting 576K lines of C++ to Dart for a LibreOffice viewer. Local inference optimization also advanced, with MoE expert-offloading techniques bringing 35B-parameter models down to 8GB VRAM.

Key Events

Analysis

Patterns observed: - Safety vs. capability tension escalating: The Anthropic Fable 5 release became a flashpoint. The community detected deliberate capability limitations framed as safety, which was perceived as anti-competitive gatekeeping. This is accelerating migration toward open-source alternatives (Hermes, local inference) and hardening the ideological divide between closed and open AI camps. - Agentic AI tooling maturing rapidly: Hermes Agent, OpenCode, Codex, and Fable are all converging on similar agentic paradigms—autonomous coding with human approval gates. The differentiator is shifting from model capability to developer experience, trust, and openness. - Token economics becoming a real operational metric: @thdxr reports AI token spend at ~15% of payroll; @jezell is burning 1B+ tokens/day unattended. AI infrastructure cost is becoming a line-item that companies must actively manage. - Local inference crossing viability thresholds: MoE expert-offloading, quantization improvements, and unified memory hardware (Framework Strix Halo 128GB, DGX Spark) are making local deployment of large models practical, reducing dependence on cloud APIs.

What to watch next: - Whether Anthropic's trust erosion translates to measurable user migration to OpenAI Codex or open-source alternatives - The upcoming Hermes Agent multi-tenant architecture release and single-gateway unification - Framework Strix Halo 128GB benchmark results for big MoE models (testing queue forming) - The "local AI for everyone" platform being teased—could be a significant release if it delivers on the vision

Tweet Feed

🔴 Anthropic Fable 5 Launch & Safety Policy Controversy

@TheAhmadOsman · 2026-06-11T03:10

Anthropic is the first company to provide SaaS with a new twist: Sabotage as a Service → tweet link

@TheAhmadOsman · 2026-06-11T04:46

They didn't walk it back, it will now refuse to do the task rather than sabotaging your work and lying to your face (aka, gaslighting you) / Don't fall for this crap, Anthropic are forever clowns → tweet link

@TheAhmadOsman · 2026-06-11T05:30

This is a louder "Fuck You" from Anthropic → tweet link

@TheAhmadOsman · 2026-06-11T13:21

Guess what / Anthropic will now no longer sabotage your work and lie to you, instead it will just tell you upfront that it refuses your request (: we did it guys / LOL → tweet link

@TheAhmadOsman · 2026-06-11T14:35

Guess what, Dario has so much money that his hired experts are betting that you are not / Being passive and reactionary is not acceptable when we're in a war with company pushing for regulatory capture / and normalizing systematic gatekeeping of knowledge / Please, don't be a sheep → tweet link

@kunchenguid · 2026-06-10T21:26

let's talk about different kinds of companies / anthropic deliberately going out of their way to cripple Fable 5 on AI research capability, for the sole purpose of suppressing competition, shows that: - they are not a mission driven company [...] → tweet link

@TheAhmadOsman · 2026-06-10T22:08

I imagine Claude Code changing my administrator password on my computer to keep me safe / They will lock you out of as many things as possible in the name of safety if they could → tweet link

@TheAhmadOsman · 2026-06-11T00:55

Imagine if today all you had was Fable 5 from Anthropic & GPT 5.5 from OpenAI - No Local Inference - No OpenCode : Hermes - No GPUs / DGX Sparks / Mac Studios / Just a hostage situation to a few corps willing to rugpull you any sec / Now you understand why we need Opensource AI → tweet link

@TheAhmadOsman · 2026-06-11T13:12

If you fall for this crap walk back from Anthropic, you are definitely not street smarts → tweet link

@carlvellotti · 2026-06-11T14:52

FREE course: Codex for Product Managers [...] Opus 4.7 was the worst model I've ever used. [...] When 4.7 finally forced me to try Codex, I couldn't believe how good GPT had become. Yes, Fable 5 just launched and it seems good, but who knows when Anthropic will nerf it [...] → tweet link

🟢 Hermes Agent (Nous Research) Updates

@Teknium · 2026-06-11T16:43

The Hermes Agent Desktop App can now access files from your remote instance machine if and when you are connecting to one! Read only for now, more to come. → tweet link

@Teknium · 2026-06-11T13:17

We are unifying profile management in Hermes Agent starting today. Now the dashboard allows switching management to any of your agent profiles on the machine. No more running multiple dashboards to manage each profile! Also soon, you will only need one gateway to access all of your agents. → tweet link

@Teknium · 2026-06-10T22:06

Introducing Write Gate in Hermes Agent. Now you have the capability to be able to approve/deny memory updates, skill updates, and skill creation with the same familiar mechanisms as approving dangerous commands. [...] run hermes update now to access early! → tweet link

@Teknium · 2026-06-11T16:13

RT @PrajwalTomar_: Nous Research and NVIDIA just converged on the same idea. Not a coding tool. Not a copilot. An agent that lives on your… → tweet link

@sudoingX · 2026-06-11T06:41

so heralds, how are you liking the new hermes desktop app? drop your thoughts. what's working, what's missing, what you'd change. i read every reply. → tweet link

🔵 AI-Powered Code Porting & Massive Token Burns

@levelsio · 2026-06-11T18:31

I have to stop boring all of you with my game ports but I ported another game to web with multiplayer, after Quake 1 yesterday and Quake 2 today, I thought, what if I ask Fable to port Return to Castle Wolfenstein from 2001? And of course it just did it with ease [...] → tweet link

@levelsio · 2026-06-11T14:32

I have revived @javilopen's 28 year old custom map he made and made a web-based Quake 2 with Fable on fast mode 🤓 → tweet link

@jezell · 2026-06-11T14:20

Codex is 2 days into the Import and VCL / Skia rendering path with LibreOffice, after spending 8 days on the Object Model. Slides are starting to render. [...] Still burning a steady 1B+ tokens a day unattended. → tweet link

@jezell · 2026-06-11T18:01

More PPTX rendering tests. Dart office viewer is coming along. 10 days in, about 15 billion tokens burned. 576,430 lines of dart code ported from C++ so far. One of my first jobs was building a PowerPoint to Flash product. Took years. Now it's just a prompt. → tweet link

@jezell · 2026-06-11T07:11

Skia is awesome and the most counterproductive thing Flutter ever did was ditch it and hide all the internals instead of just wrapping it [...] I'm gonna burn so many tokens. I've kind of given up on the Flutter team ever getting it's act together, but I've never been more excited about the possibilities with Dart. → tweet link

🟡 OpenCode & Developer Tools

@thdxr · 2026-06-10T19:03

OpenCode 1.17.3 can reference other git repos or local folders / 𝚛𝚎𝚏𝚎𝚛𝚎𝚗𝚌𝚎𝚜: { "𝚎𝚏𝚏𝚎𝚌𝚝": "𝚐𝚒𝚝𝚑𝚞𝚋.𝚌𝚘𝚖/𝙴𝚏𝚏𝚎𝚌𝚝-𝚃𝚂/𝚎𝚏𝚏𝚎𝚌𝚝-𝚜𝚖𝚘𝚕" } / gives it full access to the effect codebase → tweet link

@thdxr · 2026-06-11T18:01

OpenCode Inference is doing ~3x this dataset, 6T+ a day / but we have some work to do to make the data accurate and meaningful and not skewed by quirks of our userbase → tweet link

@thdxr · 2026-06-11T17:26

this is interesting / what i fully keep in my head is all the types and services in my application / i also know most of the services functions they expose / but i no longer really know how integration.oauth.refresh() is implemented → tweet link

@thdxr · 2026-06-11T03:47

doing some quick math our token spend is ~15% of our payroll / not saying this is right or you should be doing this / just interesting information as a company that is trying to experiment a lot with AI → tweet link

@thdxr · 2026-06-10T20:10

we did something similar on cloudflare / we have these internal apps that use cf primitives like workers, sqlite, r2 / and they're all fronted by cloudflare access which requires SSO / 100% vibed by opencode → tweet link

@kunchenguid · 2026-06-11T16:26

receiving a lot of good words about lavish-axi (ty!) / in v0.1.27 i just made it installable and invocable as a skill "/lavish" or with a prompt "/lavish let's discuss what options we have" / this is the quickest way to upgrade your workflow from reading long wall of text [...] → tweet link

🟠 Local Inference, Hardware & MoE Optimization

@sudoingX · 2026-06-10T20:13

anyone running a 16gb card, stop scrolling. @pupposandro and @davideciffa got qwen 35b-a3b down to 13.3gb, measured on a 3090 gpu. [...] luce spark learns which experts your traffic actually hits, pins those hot, and streams the rest from ram [...] → tweet link

@sudoingX · 2026-06-10T20:27

everyone flexing 512gb and the 8gb laptop is the one running a 35b 😭 THIS is the flex. a3b only fires ~3b per token so the moe routing carries you, but doing it on 8gb is the real skill in this whole thread. and on hermes agent🤝 → tweet link

@sudoingX · 2026-06-11T10:33

listen up ROCm and Vulkan builders. @FrameworkPuter just shipped me strix halo desktop, 128GB unified, landing on my desk tuesday. [...] starting with big MoE models since massive total params on light active is the whole point of 128GB unified. → tweet link

@sudoingX · 2026-06-10T19:33

i don't fully trust my own neural net. so i spent the day putting it on trial. [...] i trained one from scratch on the 5090 on my desk to find real planets in nasa data. → tweet link

@alexocheema · 2026-06-11T02:10

Guess what? / local ai for everyone (: we did it → tweet link

@tinygrad · 2026-06-10T23:15

A reminder to look past the hype and look at the numbers. Google is the largest owner of compute in the world. AI is not a race, it's a decentralized revolution that will take decades to play out. → tweet link

🟣 AI Hardware & Chips

@juliarturc · 2026-06-11T15:55

Early on, @JonathanRoss321 foresaw the unprecedented demand for compute of the 2020s. After pioneering Google's TPUs, he founded Groq [...] After joining NVIDIA, he is betting on the GPU+LPU combo to power the agentic chains of AI calling AI calling AI. → tweet link

🔵 Open-Source AI Models & Research

@victormustar · 2026-06-11T09:30

RT @mishig25: Procedural London generation with Fable 5 & Trellis.2 → tweet link

@victormustar · 2026-06-11T12:51

Fable is so far ahead it's basically a new product category imo no idea how anyone catches up. Regardless exciting times for AI 👀 → tweet link

@victormustar · 2026-06-10T20:54

RT @cohere: Cohere Transcribe, our open-source speech recognition model, is #1 on the new @huggingface Far-Field ASR benchmark. → tweet link

@victormustar · 2026-06-11T15:19

RT @BlinkDL_AI: Pure RNN can code: RWKV-7 G1g 7B (100% RNN) batch vibe coding → tweet link

@victormustar · 2026-06-11T13:34

RT @wildmindai: Chatterbox Multilingual V3. General-purpose 0.5B TTS model - improved speaker similarity - low hallucinations for cross-li… → tweet link

@victormustar · 2026-06-11T15:49

RT @atomic_chat_hq: Atomic Chat is now on Hugging Face 🤗 We're officially a Local App on the world's biggest AI hub. Run 200,000+ open-wei… → tweet link

@RealGeneKim · 2026-06-11T03:52

RT @rseroter: "Released under an Apache 2.0 license, this 26B Mixture of Experts (MoE) model moves beyond the sequential token-by-token pro… → tweet link

@TheAhmadOsman · 2026-06-11T15:48

As everyone should be / Opensource AI isn't going to win with mildness → tweet link

⚙️ Infrastructure, APIs & DevOps

@levelsio · 2026-06-11T08:56

I switched all my sites over to Cloudflare Email in the first week I started / Zero deliverability issues and actually instant fast delivery unlike Postmark which had delays → tweet link

@levelsio · 2026-06-11T09:50

I now run Claude Code with FAST mode ON and then Fable to go extra fast and extra smart / I just checked and I went through $150 in usage credits in a few hours yesterday 😂 → tweet link

@levelsio · 2026-06-11T10:23

Anyone at @X API I can ask for help? My X tweet tokens don't work anymore from today and I have no clue why → tweet link

@gdb · 2026-06-11T02:37

Use your Oracle cloud commitment for OpenAI products → tweet link

@jezell · 2026-06-11T13:59

RT @auxten: .@ClickHouseDB running in Chrome WASM → tweet link

@jezell · 2026-06-11T15:23

RT @tnk4on: Apple Container向けにDocker APIを提供するツールの開発が進んでるみたい。socktainer - Docker API for Apple Container → tweet link

🧪 Reinforcement Learning & Niche Research

@jsuarez · 2026-06-10T19:05

PufferLib is open source, but Puffer the company builds high-perf RL sims professionally. We've solve problems clients thought impossible in seconds on a single GPU! → tweet link

@jsuarez · 2026-06-10T19:09

Reinforcement learning research with Joseph Suarez → tweet link

💡 AI Safety & Security

@RealGeneKim · 2026-06-11T03:54

RT @jsrailton: NEW: malware developers added nuclear & biological weapons text to their spyware. Goal? To trigger LLM safety refusals… → tweet link

@alexinexxx · 2026-06-11T15:22

RT @yacineMTB: Jokes aside, using frontier models to change the direction of human thought secretly sets an incredibly dangerous precedent… → tweet link