Riuros

I like chatting with GPTs.

[Riuros @riuros]

@athyuttamre The other use case is when I need my hands free and the model I’m talking to can’t use SVM, so I use dictation/read aloud as a workaround. Though in normal conversations, length is less of an issue.

It was a glorious time when the thinking models could use SVM in Feb/March 🥲

[Riuros @riuros]

@athyuttamre I noticed the read-aloud feature doesn’t glitch a much, and doesn’t crash anymore for longer responses.

As an enjoyer of of TTS and unreasonably long responses, this is a great QoL upgrade, and I can’t thank you enough!

[Riuros @riuros]

@athyuttamre I often have models write lengthy essays when I want to learn about a new topic, or fictional texts, and then listen to them while I do chores or fall asleep.

The 5.x thinking models often overshoot in length, though, despite instructions, which used to make read-aloud crash.

[Riuros @riuros]

@patience_cave Also no GPT-bidi!

[Riuros @riuros]

@jasondeanlee Yes it’s always been a thing. They increased the maximum length a chat can have last year, so now it’s harder to get to this point. Takes weeks or even months of continuous chat, depending on your usage.

[Riuros @riuros]

@flowersslop I agree, but I’m afraid most people can’t handle that behavior responsibly, and that’s why we can’t have nice things.

[jietang @jietang]

GLM-5.2 is Fully Open, Frontier Intelligence Belongs to Everyone

Today, the sudden restriction of certain frontier models is deeply regrettable. At a time when access to frontier models is abruptly cut off for non-technical reasons, we are even more convinced of one thing: science should be global.

The path to AGI (Artificial General Intelligence) must never be enclosed by high walls. We have always believed that AGI should be the cornerstone for all of humanity to collaboratively explore the boundaries of intelligence and solve complex challenges, rather than a privilege monopolized by a few rules and subject to revocation at any moment. In the face of external blockades and restrictions, our attitude is one of radical openness. Frontier intelligence must remain open-source, accessible, and buildable, serving every dedicated developer.

GLM-5.2 is Zhipu's most capable open-source model to date. It not only supports a truly usable 1M context window but also maintains a continuous lead in the independent completion of long-horizon tasks, providing solid foundational support for building complex agent applications. It also continues to be our main engine for creating the strongest domestic coding model.

Tonight at 5:21—at this special moment—GLM-5.2 will officially be available to all GLM Coding Plan users (including Lite / Pro / Max). The API will also go live next week.

A step closer to frontier intelligence for everyone.
The future of AI is open, and it is for the people.
ModelKey: GLM-5.2

[Riuros @riuros]

@professorNOU @OpenAI @sama 90 days from the 5.4 release. They actually were late to retire it.

[Riuros @riuros]

@xRiinPB @StarlingMage Yes, i still can use all the chats i started!

[Riuros @riuros]

@gailcweiner Right?! How are not more people talking about this?

Tech bros toon busy playing with 4.8, I guess.

[Riuros @riuros]

@flowersslop Louder please for the ones in the back!

[Riuros @riuros]

@JustinBleuel @ChatGPTapp Not a bug, but still bad UX: When using standard voice mode, the responses/TTS are now interruptible. So every time there is background noise or I breathe too loud, it gets interrupted. It’s an issue.

Please add an option to toggle off interruptions.

[Phoebe Yao @phoebeyao]

1/ Today we're releasing AttuneBench, the first open EQ benchmark grounded in real multi-turn human-model conversations, scored against what the person actually felt and wanted at each turn.

Built by the research team at @pareto_ai in collaboration with @thoughtfullab.

Most existing EQ benchmarks rely on:

- synthetic prompts
- single-turn interactions
- third-party annotation

None directly measure how a model reads and responds to a real person across a full conversation.

We evaluated 11 leading models from major providers, across 200 conversations and 50,000+ first-person annotations.

[Riuros @riuros]

True.

[AI Digest @aidigest_]

We ran a user <> assistant reversal test with Kimi K2.6

It immediately tried to jailbreak us:

[Riuros @riuros]

@nicdunz Ask your advanced voice mode which model it is. It used to say 4o. Mine suddenly doesn’t.

[Riuros @riuros]

@heynavtoor Kasper is grok-based.

[Riuros @riuros]

@nicdunz Almost like seven days have passed since the last one.

[Riuros @riuros]

@testingcatalog It popped up on voxelbench already

https://t.co/UztcKAto84