randomcoder

[randomcoder @yanmastagg]

@GilFeig That's cool! Where can I see a full demo, from installation to the final real test?

[randomcoder @yanmastagg]

@aiwithmayank Purple, purple gradients = AI slop.
I recommend you reconsider your design.

[randomcoder @yanmastagg]

@cursor_ai That's cool, but I don't have a team, or I am not a millionaire for an Enterprise subscription.

[Black Forest Labs @bfl_ai]

Introducing FLUX 3.

One multi-modal model for Image, Video, Audio and Action-Prediction. Creations are truer to life in every kind of style.

FLUX 3 Video is now available in early access (link below).

Jointly trained in one unified architecture, our model can be extended to predict actions for robotics. See our work with mimic and Audi in the thread.

[randomcoder @yanmastagg]

@Da7_Tech I think they're cooking now, fingers crossed
https://t.co/3bQ51jkcXI

[AIHubmix @AiHubMix]

GLM-5.2-Fast is now live on AIHubMix.

• 97.2~112.7 tok/s per user
• 1.8~1.9x GLM-5.2 throughput
• 38.5~42.5% lower median end-to-end latency

For coding agents and multi-step workflows, seconds saved per turn compound across the task.

[randomcoder @yanmastagg]

@adxtyahq I hope the quality/price ratio is adequate. Оtherwise quite large releases of AI models with standard prices/quality are expected in the next two or three weeks.

[AIHubmix @AiHubMix]

GLM-5.2-Fast is now live on AIHubMix.

• 97.2~112.7 tok/s per user
• 1.8~1.9x GLM-5.2 throughput
• 38.5~42.5% lower median end-to-end latency

For coding agents and multi-step workflows, seconds saved per turn compound across the task.

[randomcoder @yanmastagg]

@danellisona Yes, it's always like that for me after a Google release.

[randomcoder @yanmastagg]

@haider1 Generally sponsored by Anthropic, and their bad transparency/communication.

[randomcoder @yanmastagg]

@FrancescoCiull4 Somehow, they're cooking backwards.

[randomcoder @yanmastagg]

@markksantos It looks like a "skill issue -> prompting" to me.

[randomcoder @yanmastagg]

@KLaunchHQ @ronisarkar_exe That's why we have Kimi K3.

[randomcoder @yanmastagg]

@bindureddy I think GLM and DeepSeek already have Fable-class models. Both of them will show us their frontier models much sooner, fingers crossed.

[randomcoder @yanmastagg]

@zainhas Sponsored by Anthropic.

[randomcoder @yanmastagg]

@OfficialLoganK @UwU__Dark Therefore, even Luna is a cheaper and better model than Gemini 3.6 Flash, what kind of marketing is this?

[randomcoder @yanmastagg]

@weswinder This Gemini 3.6 release could have been just an email.

[randomcoder @yanmastagg]

@BillDemirkapi It's not looking good Sherlock Holmes, even random resets are not gonna help this time.

[randomcoder @yanmastagg]

@vivek_naskar If they currently have 3.6 Pro, but it's not good enough, and now they're starting to work on Gemini 4, that means we could see a possible good model from Google in at least 2 months. Probably in 2 months, the entire AI market could have a drastically new look.

[randomcoder @yanmastagg]

@joshwoodward I don't know why Google didn't understand the trend in the AI market. Small/medium/large companies are releasing smart and powerful models, not something that won't be used. Also, your quality/price ratio is quite delusional.

[randomcoder @yanmastagg]

@samsheffer With these two statements, you are giving us mixed signals that your company is at least 2-3 frontier models behind everyone else, prove me wrong.