Tag: xai

xAI launches Grok 4.1; faster and less likely to hallucinate

(Picture: generated)
In stealthily rolling out the latest Grok iteration, xAI found ~65% of users preferred the new bot, which also currently tops the leaderboard for Lmarena – text with about ten points.

— Our 4.1 model is exceptionally capable in creative, emotional, and collaborative interactions, xAI writes on their launch page, and adds:

— It is more perceptive to nuanced intent, compelling to speak with, and coherent in personality, while fully retaining the razor-sharp intelligence and reliability of its predecessors.

This is pursuant to the launch of GPT-5.1 last week; labs are working a lot more on personality for their bots, and want their chatbots to be good at actually chatting.

In addition, the new bot boasts of far lower hallucinations, dropping from 12% for Grok 4 to 4.22% for 4.1, and should also be a lot faster.

The Verge notes that the bot seems to carry on the laissez-faire rules of its predecessors, opening up to NSFW use cases, although it hasn’t gone full mecha-Hitler yet.

Read more: xAI’s launch page, note on The Verge.

xAI releases Grok 4 Fast, focusing on cost and efficiency

Grok 4 Fast is cheaper and uses fewer tokens, xAI says.
Almost as good as Grok 4, xAI claims — and plenty cheaper. Meet Grok 4 Fast. (Picture: xAI)
Touted as a state-of-the-art model for less, they claim to be «pushing the boundaries for smaller and faster AI.»

They say the model achieves performance comparable to Grok 4 proper with 40% less token use, and even pushes the price of those tokens down by 98%.

Continue reading “xAI releases Grok 4 Fast, focusing on cost and efficiency”

Friday roundup: A good week for coding, speech models

Coding and speech models grab the headlines for this weeks roundup.
Both OpenAI and Microsoft are out with speech-to-speech models this week. (Picture: OpenAI)
OpenAI makes Realtime API generally available
The agentic Realtime model is a native speech-to-speech model that can be used to make customer service agents, phone reps and voice navigation features. It doesn’t go through speech-to-text and text-to-speech loops and generates audio «directly through a single model and API.» OpenAI is marketing this to developers who want more natural flowing speech, and it’s not available as distinct model in ChatGPT – yet. You can hear it and see it in use at places like Zillow, T-mobile, StubHub and Oscar Health, though. With general availability, it will surely show up in a lot more places soon.
More at: OpenAI’s launch page, discussion on r/OpenAI.

Read on for more news!

Continue reading “Friday roundup: A good week for coding, speech models”

X.ai open sources Grok 2, now available on Hugging Face

Grok 2 goes open source, with Grok 3 to follow in six months.Elon Musk just announced on x.com that the model has been released as «open source» under a «Grok 2 Community License Agreement.»

The license forbids the use of the material to «train, create or improve» any foundational AI models, but does allow for some commercial use.

Musk also said that Grok 3 — the model currently running on x.com, will be open sourced in about 6 months:

Continue reading “X.ai open sources Grok 2, now available on Hugging Face”

Short Friday news; Meta’s bias checker, how to deal with children, Gemini’s new memory, and more

Metas new bias-chief is a right wing influencer who crusaded against DEI politicies
Generated picture.
Meta’s new «bias»-checker is a right-wing influencer
Robby Starbuck rose to fame as an influencer campaigning against Diversity, Equity and Inclusion (DEI) in the USA. He often sued companies to force them to end such policies, and even sued Meta after their AI wrongly implicated him in the events of January 6.

Now the suit is settled and he has a new job offer; as a Meta advisor to address «ideological and political bias» in their AIs. This is what Trump meant when he went against «woke» AI, and Meta says they have made «tremendous strides to improve the accuracy of Meta AI and mitigate ideological and political bias» since working with Starbuck.
More at The WSJ, The Verge, Mashable and MSNBC.

Gemini now defaults to remembering previous chats
Google Gemini’s new feature is always on by design, and will remember your older chats without specifically asking. The feature delivers «more personalized responses the more you use it,» Google says. It will remember «key details and preferences you’ve shared, leading to more natural and relevant conversations, as if you’re collaborating with a partner who’s already up to speed.»

It can be turned off by going to Settings, then «Personal context.» There is also an option called «Temporary Chats» that won’t be remembered.
More at Google’s launch post, The Verge and 9to5Google.

Continue reading “Short Friday news; Meta’s bias checker, how to deal with children, Gemini’s new memory, and more”

So, who really has the most GPUs in the AI race?

AI companies are investing greatly in GPUs and its hard to catch up. Here's what we know so far.
According to this chart, the best compute units are GPUs, by far — and those green dots are all made by Nvidia. (Picture: Steve Jurvetson, CC BY 2.0)
Sam Altman says OpenAI will deploy more than a million GPUs by the end of the year. Meta says it’ll beat that. Elon says xAI is building the world’s biggest AI supercomputer.

But here’s the truth: None of these numbers are easy to verify — and each company counts differently.

Continue reading “So, who really has the most GPUs in the AI race?”

Grok’s new «companions:» sex crazed lovebot and a profane firestarter

Profanity or naughty, Groks new "companion" looks for your seedy underbelly -- not aloof phiosophy.
Looking at the seedier underbelly of the Internet for inspiration? NSFW and profanity laden animated avatars lead the way for Grok.
Just a few hours after Elon Musk’s Grok chatbot launched two AI animated avatars, the Internet had a chance to test them out.

One is a sexy anime waifu called Ani, that «is obsessed» with you and totally in love. The bot gets more and more NSFW as you progress and «level up» to lingerie views and increasingly raunchy chats, writes TechCrunch.

Continue reading “Grok’s new «companions:» sex crazed lovebot and a profane firestarter”