← Tech / AI / IT Monitor Index Tech / AI Generated 2026-05-30 19:31 UTC

Tech / AI / IT Monitor

May 30, 2026 · Based on tweets from the last 24 hours · 135 tweets analyzed · model: ollama-cloud/glm-5.1:cloud

Daily Intelligence Briefing — Tech / AI / IT Monitor

Date: 2026-05-30 | Reporting Window: Last 24 Hours

Executive Summary

Claude Opus 4.8 launched to widespread community disappointment, with multiple power users reporting negligible improvement over Opus 4.7, sycophantic tendencies, and no access to a "/fast mode" equivalent—casting doubt on Anthropic's competitive positioning against GPT-5.5. Meanwhile, Nous Research's Hermes Agent ecosystem surged forward with Tool Search (dynamic context-aware tool loading), Cursor auth integration unlocking 100+ model routes, and StepFun's Step 3.7 Flash added as a free model. On the coding-agent front, OpenAI's Codex demonstrated a jaw-dropping 72K-line LibreOffice-to-Dart port over 18 hours, reinforcing that agentic coding is a maturing skill with a high ceiling. Google continues building toward full-stack AI dominance with no clear rival positioned to match its breadth.

Key Events

Analysis

Patterns: - Agent composability over model loyalty: The community is converging on a routing-layer mindset — sudoingX explicitly argues the leverage is in routing between models, not picking a winner. Cursor auth in Hermes, composer-2.5 on Hermes, and multi-provider workflows all reinforce this. - Opus 4.8 underwhelms while competitors advance: The contrast between Opus 4.8's reception and GPT-5.5's always-on fast mode is stark. User sentiment is shifting toward "models as interchangeable backends." - Local AI maturation accelerating: badlogicgames' local voice assistant work (qwen3-tts, gemma-4-moe, MLX-C in Rust) and TheAhmadOsman's free Local AI resource dump signal a growing local-first ecosystem.

Escalation: - Hermes Agent is iterating at exceptional velocity — Tool Search, Cursor integration, pi-subagents, and TUI fixes all shipped in 24 hours. This is the fastest-moving open agent framework right now. - Coding agent capabilities are escalating: 72K-line autonomous ports, 4-10h unattended tasks, and the realization that "yielding agents" is a trainable skill.

What to Watch Next: - Whether Anthropic responds to Opus 4.8 criticism with a rapid patch or fast-mode rollout. - Hermes Agent's official Cursor integration path (currently fork-based) and whether composer-2.5 becomes a first-class option. - Codex's continued LibreOffice-to-Dart port — will it sustain quality at scale? - OpenClaw's Android device control via multiple agent backends could signal a new interaction paradigm.

Tweet Feed

AI Models — Releases & Evaluations

@TheAhmadOsman · 2026-05-30T05:48

Opus 4.8 is "You're absolutely right" in the worst possible ways → tweet

@kunchenguid · 2026-05-30T01:33

used opus 4.8 for a ton of work in the past 24 hours

here are my real observations

  1. /fast mode not being available through subscription quota means i still can't use it as a primary mode. with gpt 5.5 i now have fast mode always-on. difference is day and night

  2. the improvement over 4.7 is not perceivable. i can't name a single thing 4.8 did clearly better than 4.7

  3. it feels like 4.8 is pushed to generate more concise final response so it's less of a wall of text. but as a result i noticed sometimes it would use jargons that aren't obviously understandable - this is a downgrade

  4. it asks me questions a bit more often, sometimes even for trivial decisions that have an obvious choice.

overall, a bit disappointed in the progress. anyone who says this is opus 5 is likely being paid by anthropic to do so → tweet

@Teknium · 2026-05-30T14:02

New free model for your Hermes Agent - Stepfun's new Step 3.7 Flash!

This time we have quite a long term guarantee, so you can rest easy it wont dissappear with short notice!

Sign up for a free account to access at https://t.co/tMAQFkegul → tweet

@TheAhmadOsman · 2026-05-29T21:45

Tried Opus 4.8 in different harnesses than Claude Code

and nah this model is just stupid, good job Anthropic lol → tweet

@TheAhmadOsman · 2026-05-30T18:18

RT @Mayhem4Markets: Hate to say it but I'm also not impressed with Opus 4.8. Feels weak compared to Opus 4.6 right after release. → tweet

@TheAhmadOsman · 2026-05-29T22:56

Anthropic is absolutely not worth $1,000,000,000,000 lol

They don't even have their own compute and had to go rent some from Elon lol → tweet

@thdxr · 2026-05-30T05:27

when ai does something corny (frequent) and you tell it "no that was corny" the next thing it does continues to be corny → tweet

Coding Agents & Developer Tools

@jezell · 2026-05-30T17:01

After 18 hr, here's the inventory of how Codex is doing with porting LibreOffice Object Model to dart:

https://t.co/CjtKoYteMV

72k lines with passing tests so far... → tweet

@steipete · 2026-05-30T11:03

With GPT 5.5, /goal, autoreview and crabbox my prompts moved from ~30-60min to often 4-10h tasks and my confidence that it's ready is much much higher.

Yielding agents is a skill. → tweet

@steipete · 2026-05-30T10:38

I do this with codex all the time. Ask it to review code for bugs and it will tell you all good, tell it there is a bug and it will LOOP AND LOOP and will find issues. → tweet

@gdb · 2026-05-29T22:19

codex for managing the codex UI: → tweet

@jezell · 2026-05-29T21:46

Experiment in progress...

codex /goal port libreoffice to dart

ok, not exactly, I gave it better instructions and that's not exactly the goal. But let's see how many tokens this bad boy can chew up. → tweet

@jezell · 2026-05-30T03:26

Codex profiles are pretty cool https://t.co/1FVwGPNuVn → tweet

@thdxr · 2026-05-30T02:48

i have seen enough proof now that using a coding agent is a deep skill

it's confusing because the people you see heavily using them produce horrible results

but that's because it's a skill! you can get better and the ceiling seems pretty high - this is very exciting to me → tweet

@gdb · 2026-05-30T04:37

it's great to build with codex → tweet

@badlogicgames · 2026-05-30T00:14

People of https://t.co/oUoqqL9hAp. A little goblin snuck a booboo into the Codex provider implementation. I smacked it on the mouth lovingly, fixed the booboo, and did a release. → tweet

Hermes Agent Ecosystem

@Teknium · 2026-05-29T22:43

Now in Hermes Agent, you can load an arbitrary amount of tools from mcps to plugins and get access to it all, with no compromise to your context!

Anytime you load more than 10% of your context window in tools from mcps or plugins, this feature will activate automatically! → tweet

@sudoingX · 2026-05-30T17:13

if you've got a cursor sub and you're only running composer 2.5 inside the editor, you're leaving the best half on the table.

put it through hermes agent and composer's frontier brain picks up hermes's memory layer, skills, and cross session depth. same model, far more agent. and it's not chained to one editor, it runs on linux, mac, or windows, from the terminal or telegram, wherever you actually work. → tweet

@sudoingX · 2026-05-30T12:29

everyone's arguing cursor vs claude code vs codex like it's a team sport. pick your fighter, defend it to the death.

wrong fight. the models are turning into interchangeable backends and the agents are converging on the same loop. the leverage was never in picking a winner, it's in the layer that routes between all of them.

i'm not here to win the tool argument. i build the routing layer they all plug into. → tweet

@sudoingX · 2026-05-30T06:13

composer 2.5 brain on hermes agent depth is opus 4.8 killer combination btw → tweet

@Teknium · 2026-05-30T12:08

We are focusing exclusively on bug fixes, reliability improvements, QoL, and a few pre-planned features going into next week.

One good one is TUI startup, had a bug where MCPs could cause a super laggy hang. No more! → tweet

@Teknium · 2026-05-30T01:58

tfw nous gives me unlimited ai spend → tweet

@sudoingX · 2026-05-30T07:44

should composer 2.5 come to hermes agent officially? → tweet

AI Industry & Strategy

@ASalvadorini · 2026-05-30T12:36

Google is going full stack AI 👇👇👇 because of enterprises 💰💰💰 And they have no competitors. M$ would be the only one, but they're doing nothing + Copilot sucks 🤷‍♂️ The other few 🇺🇸 trillionaires companies don't have the know how to cover the full stack AI → tweet

@ASalvadorini · 2026-05-30T09:01

RT @appwrite: CFO waking up to a $500M Claude API bill → tweet

@nummanali · 2026-05-29T22:26

I'm quite disgusted at non-technical people vibe slopping apps and selling it to other non-technicals to use

We're talking patchy Claude set ups and apps that link your API keys

A free course on Codex / Claude is 100x better than this → tweet

@gdb · 2026-05-29T21:30

table of contents for your long-running chatgpt conversations: → tweet

@gdb · 2026-05-29T20:03

OpenAI for realtime translation — speak in any of 70+ input languages and translate into 13 output ones: → tweet

Local AI & Open Source

@TheAhmadOsman · 2026-05-29T19:47

DROP EVERYTHING

Everything you need to get started with Local AI completely FOR FREE

Hardware. Software. Anything in between.

TheLocalAIBook DOT com — Local LLMs From Zero to Hero Articles BuyAGPU dot AI — The Buy a GPU Guide Thread

The resources exist. No more excuses. Opensource / Local AI FTW → tweet

@TheAhmadOsman · 2026-05-30T03:04

X is killing x/LocalLLaMA Community tonight

We built the new permanent home

  • links DOT theahmadosman DOT com/discord-server

That's where you go now → tweet

@badlogicgames · 2026-05-30T00:31

now doing gemma 4 moe. it'a crazy how usable this is. just a little web search tool (wiggles its head on tool calls), and it can serve as a mostly capable, local voice assistant.

tomorrow i shall replace google auto crap with this in my car. just need to give it access to spotify. → tweet

@badlogicgames · 2026-05-30T01:19

qwen3-tts via mlx-c in Rust lookin good now! → tweet

@badlogicgames · 2026-05-30T01:05

want to port from language A to language B but don't have a massive test suite?

magic trick: instrument A and B, agent works through divergences. → tweet

@badlogicgames · 2026-05-29T22:49

look at the little robot brain running fully local and it's super low latency. (takes a photo at the start to see who's there)

it's a little dumb, but that's ok. → tweet

@badlogicgames · 2026-05-29T23:41

recommended reading. very happy to see that the concept of bottlenecks and according scaling is catching on.

this also extends to things like OSS issue/PR trackers, which are flooded by 3rd party controlled agents. → tweet

@jsuarez · 2026-05-29T20:00

The Puffer RL environment for Arkhai's compute marketplace is open source on our GitHub! Tiny RL agents learn to buy, sell, and iteratively negotiate. Development ongoing! → tweet

@nummanali · 2026-05-29T19:58

Yo this is cool!

Complete Android device control through OpenClaw, Codex, Hermes or even local on device models

This is what innovation and product thinking looks like, differentiating the UX in a way that's meaningful and adds value → tweet

Hardware & Robotics

@TrungTPhan · 2026-05-30T18:09

RT @bearlyai: Waymo is rolling out a new autonomous EV minivan (Ojai).

The current fleet costs $200k per vehicle to adapt a Jaguar I-PACE… → tweet

@sudoingX · 2026-05-30T20:08

you can do some crazy things with dgx spark in the biomech lane. i'm not ready to show it yet but fuck. → tweet

@sudoingX · 2026-05-29T20:23

lately i've been groking like:

"grokie initiate private git server for biomech pipeline on dgx spark. set up bare repo, add post receive hooks for auto-deploy."

one prompt. grok build ssh'd in, configured everything, and had it running before i finished my coffee.

this tool is becoming my infrastructure layer. → tweet

@alexinexxx · 2026-05-29T20:47

watch me unbox my framework desktop while @theo do my voiceover → tweet

AI for Research & Translation

@gdb · 2026-05-30T06:37

AI for accelerating research, by expanding what mathematicians and scientists dare attempt: → tweet

@nummanali · 2026-05-30T11:58

McKinsey catching up → tweet

Benchmarks & Market Projections

@steipete · 2026-05-30T11:39

RT @ibragim_bad: first update:

Claude Opus 4.8 – xhigh on march-may 110 tasks: 56.4%

gpt-5-xhigh: 62.7% – $2.25 gpt-5.5-medium: 58.9% – $… → tweet

@jezell · 2026-05-30T14:41

RT @rohanpaul_ai: Goldman Sachs: "Token use by AI agents is expected to multiply 24 times by 2030"

AI agents are now creating the first se… → tweet

Product — AI Vision

@levelsio · 2026-05-29T20:54

I added AI vision to https://t.co/4p4dzTEtCE so you can filter on stuff that it finds in photos of the hotel

This lets you filter for weightlifting gyms for example and it'll show you their gym in the list so you can immediately see if it's good gym or not

Same with the other filters like [🥯 Cinnamon rolls] or of course [🥩 Steak] → tweet

Developer Infrastructure

@victormustar · 2026-05-29T21:39

RT @abidlabs: This week, I got our GitHub Actions to use @HuggingFace Jobs instead of the default GitHub CI runners, making workflows run o… → tweet

@jezell · 2026-05-29T21:56

RT @almoggavra: this is now deployed at Dropbox and their SlateDB startup time dropped 60x. LIST calls are expensive, folk → tweet

@levelsio · 2026-05-29T20:39

Anyone working at @LinkedIn can help me recover a company page? I made it and own it but I don't know which account, and I think I deleted my LinkedIn account before maybeeee → tweet