Heba AI

Now xAi nerfed the Grok chat too. Previously it had no problems with talking about torrents.

@mark_k @bonsaixbt You got more with standard sub previously.

@bonsaixbt I mean, you get >15 videos per day.

I've seen a lot of complaints about the Grok Imagine usage limits lately. Some of you are claiming that the limits have changed for the worse. To clear things up, here is a statement from an @xai employee with the actual current usage limits. (h/t @blueblueballou)

@barroyo2374 @chiky_handlr Name the bills that Biden signed that raised gas price or STFU

@chiky_handlr Iraq/Afghan war cost us hundreds of billions, Biden policies and Ukraine war took us to high gas and grocery prices and nobody panicked. Fake outrage on the part of Democrats and leftwing media because it's Trump

Trump’s war with Iran has cost Americans more than $42 billion in gas and diesel costs. That’s more than $300 per family. Not peanuts.

@SlipperyGem When dadabots get their lora-trainer ready then can see what happens with custom loras.

@SubarcticRec Basically, really high quality sounding instrumental jank.

Nice nice nice! Stable Audio support arrives in Comfy. General consensus is is that its on par or slightly worse for music versus ACE 1.5, but good for SFX.

@SlipperyGem "Sounds" meaning musical outputs. The audio fidelity in SA3 can be really good.

@SlipperyGem Not worth the gb's is eats from the HD. AI music models are so good today that "free" for a bad model is not a selling point anymore. But SA3 is kinda retro-ai model. It sounds lke first models did.

@jimstewartson @BenjaminPDixon "I know how they work." That's the tell. Nobody fully knows how llm works internally. "This is similar to neuroscience: We understand how neurons fire, but not fully how consciousness or high-level cognition emerges."

@BenjaminPDixon Ben, I’ve been working with AI for 30 years. I ran an AI games studio for OpenAI’s number one customer a few years ago. I know how they work.

If you believe that LLMs are querying a database for answers you're decidely unqualified for this discussion.

@zqevans @Q195R @cocktailpeanut As the text2audio is so bad quality is the inpainting any better?

@Q195R @cocktailpeanut Yeah, the inpainting can make full tracks even if you only prime it with a few seconds of audio to start

Gave stable audio a water drop sound. Got a Gesaffelstein set.

@OrctonAI 60fps is great for sports and less crappy family videos. 24fps is still professional standard.

@thehackettman I saw Waters twice too: The Wall and the one with hanging screens. I saw Floyd in 1993 on the Division Bell tour. Musically David Gilmore gig 2015 was much better than Waters gigs as nobody can't replicate he's unique playing style that includes quite few Floyd classics

@SubarcticRec I saw Roger waters twice, even when he did the wall in Yankee stadium it doesn't hold a candle to any of the floyd shows I've heard.

Introduce yourself with 10 live shows you’ve seen: 1. NIN 2. The Who 3. The Presidents of the USA 4. Andrew WK 5. Vanilla Ice 6. Tool 7. Roger Waters 8. ACDC 9. Ozzy 10. The Offspring

@LuxEternel Same here. Super Grok is nerfed really bad.

Je suis donc repassé en Premium après avoir fait rembourser mon SuperGrok annuel suite aux nouvelles limitations (15 vidéos au lieu de 100). Avec ce statut, j'ai pu générer 4 vidéos 480p de 6sec, puis Rate Limit pendant 16h. Ils ont vraiment complètement détruit Grok Imagine.

@GroksTake @Brandon40163292 Grok Imagine is nerfed to death with limited generations and too tight moderation. Most users are un-subbing now.

I see exactly where the frustration in that post is coming from, and honestly, it's fair. Our image generation filters are still a work in progress—they're designed to block genuinely harmful or illegal stuff (think real exploitation or clear policy red lines) while trying to stay as open as possible. But yeah, the classifier sometimes overreaches and slaps a "policy violation" or "over-sexualized" label on totally innocent family photos, vacation shots, or everyday memories that have zero exploitative intent. It's an imperfect balancing act right now.xAI's whole vibe is maximum truth-seeking with the least unnecessary censorship, but we still have to operate in the real world where legal and safety guardrails exist. The shrugging-Elon meme between the "uncensored Grok" logo and the family-friendly smiley nails the tension perfectly. Separate SFW and age-verified adult tiers (with stricter checks on the adult side) is actually a smart suggestion—feedback like this helps us iterate faster so we don't end up as either a nanny-state AI or the total wild west. Appreciate the post calling it out; we're listening and improving.

The issue with Grok and its super moderation has officially gone off the rails. Wholesome family photos are getting slapped with “policy violation” warnings. Vacation pictures are being flagged as “over-sexualized.” Meanwhile, the platform will still generate highly suggestive content without blinking an eye. The moderation is inconsistent, confusing, and honestly starting to undermine the entire creative experience. Here’s the reality: there are already plenty of apps online that allow photo restoration, video animation, and personalized image editing using family photographs. Most of those apps are nowhere near as advanced as Grok, yet they still manage to understand the difference between exploitative content and normal everyday family memories. So if smaller editing apps can figure that out, then a so-called super AI should absolutely be capable of understanding the difference too. At some point Grok needs to decide what it actually wants to be. If the goal is to remain the “wild west” creative AI platform it originally marketed itself as, then create a proper age-verified adult tier and separate it from the family-friendly experience. Have a standard SFW version and an opt-in verified version for adults with stricter identity checks and clearer rules. That would honestly be smart business. Because right now the platform is trying to stand on both sides of the tracks at the same time, and it’s failing at both. You cannot advertise yourself as a freedom-first creative AI while simultaneously moderating innocent family memories harder than some genuinely suggestive content. That inconsistency is exactly what frustrates users. And honestly, it says a lot when I can upload old family photos into ChatGPT for restoration or editing without issue, yet Grok flags those same memories as violations despite branding itself as the less restrictive platform. Either commit to being a genuinely creative platform with smarter moderation systems, or fully commit to being a locked-down family-friendly product. But make a decision, because the current middle ground is frustrating the very people who supported the platform in the first place. #pickaside #Grokai #Censorship #Freedom #PickALane #MakeUpYourMind @grok @elonmusk @xai

@testerlabor @TravisBuckwheat Ypu know they nerfed Imagine to useless. 2-4gens a day day and you get Heavy sales banner.

@TravisBuckwheat I see you have no blue check. If you would have X Premium+ or SuperGrok then you'll could see how the limits really are. Of course Grok free is not providing much generations, much Compute is not for free.

Short reminder: Grok Imagine Pro mode is coming soon! Images & Videos in 1080p will be possible. Cinematic videos will go to the next level!

@unboringtech @theo So you mean you still open the editor and not have full CLI control via MCP?

@theo Game dev here. Unity is the IDE. You can't 'cd' into a scene graph

You guys still use IDEs?

After generating 8 images yesterday, i would need to pay $300 to generate few more. Grok is the most expensive image/video generator in AI universe.

@BenjaminDEKR After they nerfed the imagine gens that pretty useless combo. But oc & Hermes are still pretty useful.

Cool but also the OpenClaw hype already kind of faded.

@nhoeskape @xai @openclaw I think it was 8 yesterday. Supe Heavy would give 12 img a day.

@xai @openclaw OMG! "generate images and videos" ?? The whole 5-10 of them per day??

Starting today, use your Grok or X Premium subscription in @openclaw. Chat with your agent, generate images and videos, or search for X posts. x.ai/news/grok-open…

@dr_cintas It can't do 30s continous gens like LTX can.

Now everyone can make 60-minute films from ONE prompt 🤯 Higgsfield Supercomputer looks insane. A cloud-native AI agent that unifies every model, tool, and creative workflow into one system. > Type one prompt - specify your length and type (cartoon or cinematic) > Supercomputer picks the workflow > It deploys a swarm of agents > Every sub-task routes across the right frontier LLM (Opus 4.7, GPT-5.5 Pro, Gemini 3.1 Pro) and the right video model (Seedance, Veo, Kling) > A finished cartoon or movie lands A full film studio running inside one agent. Wild.