New robot waifus, GLM 5.2 craze, AI spas, new world models, new science agents: AI NEWS
Credibility score: 51/100 — Mixed Credibility. Several questionable claims detected. Watch with healthy skepticism.
BSmeter analyzed "New robot waifus, GLM 5.2 craze, AI spas, new world models, new science agents: AI NEWS" and rated it 51/100 for credibility (a BS score of 49/100 — mixed credibility), on 2026-06-23. Its weakest claim — "Claiming DreamXWorld is more consistent than other models like World or HY World Play." — scored 45/100 and was flagged as confidence mismatch. 28 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Of 28 claims analyzed: 0 scored under 40, 23 between 40 and 69, and 5 at 70 or above.
Claims analyzed
Video opens with a highlight reel preview of various AI advancements and products. — Just Vibes (50/100)
At 0:00
This is just a rapid-fire intro, throwing out all the buzzwords to get you hyped. Waifus, new models, local runs—it's the AI equivalent of a movie trailer. 🤖✨
Why this score: The speaker is setting the stage with a high-energy montage of upcoming topics, using exciting, attention-grabbing phrases like 'absolutely insane' and 'takes the industry by storm.' It's a classic hook, designed to build anticipation rather than make verifiable claims at this point. It's pure entertainment and hype building.
Original quote: “AI never sleeps and this week has been absolutely insane. Your full body robot waifu is coming soon. The new number one open source model GLM 5.2 takes the industry by storm and now you can even run this locally on high-end consumer devices. We have a new open source image model which can also edit…”
Video opens with a highlight reel preview of AI news, setting an excited tone. — Just Vibes (50/100)
At 0:30
Just a rapid-fire intro of all the cool AI stuff coming up — Midjourney spas and robot demos? Sign me up! 🤖🏨
Why this score: The speaker is giving a quick, high-energy preview of the topics to be covered in the video, using exciting examples like 'Midjourney turns into a spa hotel' and 'ridiculous humanoid robot demos' to hook the viewer. It's pure hype, not a claim to be fact-checked.
Original quote: “consistent. Midjourney, the image generation company, now turns into a spa hotel. Like seriously, we have an incredibly powerful and open source world generator. Alibaba releases an open unified model for science. It can understand things like proteins, small molecules, chemical reactions and more.…”
Claiming DreamXWorld is more consistent than other models like World or HY World Play. — Confidence Mismatch (45/100)
At 0:30
Says 'as you can see' but we're just seeing a demo, not a side-by-side comparison. Trust me, bro energy. 👀
Why this score: The speaker asserts that DreamXWorld is 'even more consistent' than other models, using the phrase 'as you can see.' However, the video only shows a demo of DreamXWorld, not a direct, objective comparison with the other models mentioned. This relies on the viewer accepting the claim without direct visual evidence of the comparison.
Original quote: “Especially if you compare this with other similar models like World or HY World Play, then as you can see DreamXWorld is even more consistent.”
Introduces DreamXWorld, an AI model for creating explorable, evolving worlds from prompts. — No Frame (75/100)
At 0:50
Describes DreamXWorld's core function: creating interactive, evolving environments from prompts. Sounds like a game changer. 🎮✨
Why this score: The speaker is introducing a new AI model, DreamXWorld, and describing its functionality based on its stated purpose. The description is straightforward and aligns with the typical capabilities of generative AI models for virtual environments. No obvious framing or manipulation here.
Original quote: “First up, we have a new world model called DreamXWorld. With just a prompt or reference images, this lets you create a world that can be explored, controlled and changed with actions or event prompts. Basically, it takes in instructions like movements or camera control or events and outputs this…”
Highlights DreamXWorld's training data and ability to generate long, consistent videos. — No Frame (75/100)
At 1:17
Explains the training mix and the 'persistent memory' for long, consistent videos. That's a big deal for AI generation! 🧠🎬
Why this score: The speaker is providing technical details about DreamXWorld's training methodology (Unreal Engine, gameplay data, real-world videos) and a key feature: generating long, consistent videos with persistent memory. These are verifiable technical claims about the model's architecture and performance, presented without exaggeration or loaded language.
Original quote: “For example, you can even add this drone here or ride a motorcycle like this or drive a car or do a variety of different stuff. Now, what's interesting is that it's trained on a mix of Unreal Engine and gameplay data as well as real world videos. So it learns both realistic movement and more…”
Claims DreamXWorld is more consistent than competitors and the 5B version is small enough for consumer devices. — Confidence Mismatch (45/100)
At 1:45
Says the 5B version is 'fairly tiny' at 21GB for 'consumer devices.' Bro, my consumer device is crying in 8GB RAM. 😭💾
Why this score: The speaker confidently states that the 21GB 5B version of DreamXWorld is 'fairly tiny' and 'should be able to fit this on both consumer devices.' While 21GB might be small for a large AI model, it's a significant chunk of storage and memory for many standard consumer devices, especially if they have limited RAM or older hardware. The claim of easy fit on 'both consumer devices' (implying widespread compatibility) is an overstatement without specifying hardware requirements, leading to a confidence mismatch.
Original quote: “Especially if you compare this with other similar models like World or HY World Play, then as you can see DreamXWorld is even more consistent. The awesome thing is they released this already. So if you scroll up to the top and you click on this code button and you scroll down a bit here it contains…”
Introducing Permavid to fix AI video consistency issues — No Frame (75/100)
At 2:30
Setting up the problem Permavid aims to solve in AI video editing. Sounds like a genuine innovation.
Why this score: The speaker is introducing a new AI tool and explaining the problem it addresses in AI video editing, specifically the lack of consistency in edits. This is a straightforward introduction of a new technology.
Original quote: “we have a new AI called Permavid. This is focused on one of the biggest problems in AI video editing, which is consistency. Right now when you edit a generated video, for example, adding an object or replacing an object or changing the style of the video, the model often forgets the change later…”
Claiming code released, then immediately retracting it. Classic volume game 📢 — Volume Game (45/100)
At 4:30
Says 'released a code button' then 'haven't released any code or models yet'. The ol' give-and-take in one breath. 🎭
Why this score: The speaker initially states that a 'code button' has been released, creating an impression of immediate availability. However, they immediately follow up by clarifying that no actual code or models have been released yet, effectively retracting the initial claim. This is a volume game, where a bold statement is made and then quietly walked back.
Original quote: “Now at the top, they have released a code”
Hyping up advanced video tech, then admitting no code or models are released yet. Classic 'future tech, current vaporware' move. 🤡 — Volume Game (45/100)
At 4:30
Talks up all these 'cool' effects, then drops the 'no code yet' bomb. The hype is doing overtime for something that doesn't exist publicly. 😬
Why this score: The speaker builds significant excitement around the capabilities of this new video generation tech, listing advanced techniques it can handle. Immediately after, they reveal that no code or models have actually been released, effectively retracting the immediate utility of the impressive claims. This creates a 'volume game' where the initial, louder claim of capability is undercut by a quieter, later admission of unavailability.
Original quote: “The really cool thing is that this can even handle special techniques like dolly zooms or bullet time effects or lens distortions. So here's a bullet time effect and then here's a dolly zoom effect and here's some really wide fish eye distortion. It's able to understand this and apply it to the new…”
Claims their model outperforms competitors on their own leaderboard. Shocker! 🙄 — Confidence Mismatch (45/100)
At 6:30
They're citing their own leaderboard as proof of superiority. That's like a kid grading their own homework. 📝
Why this score: The claim of outperforming competitors is made based on their internal 'Boo Goo Arena' leaderboard. While not necessarily false, it lacks independent verification, making it a self-serving claim of superiority without external validation. It's a classic move to boost confidence without external proof.
Original quote: “Here if you look at their leaderboard called Boo Goo Arena, then here they claim that it does outperform existing open source image generators like Z image or Kwen or Hi Dream.”
Announcing a 'really useful' AI model for science from Alibaba's Tongyi Lab. — Loaded Language (45/100)
At 8:30
Calling it 'really useful' before any details. That's some pre-hype, baby! 🚀
Why this score: The speaker uses the phrase 'really useful' to describe the AI model before providing any specific information about its capabilities or applications. This is loaded language designed to create a positive impression without objective evidence yet.
Original quote: “Also this week, the Tongyi Lab from Alibaba releases a really useful AI model for science.”
Logos models are more performant than competitors and run on consumer devices. — Confidence Mismatch (45/100)
At 10:30
Claims 'a lot more performant' and 'most consumer devices' without showing the benchmarks or defining 'most'. That's a lot of confidence for zero visual proof. 📈👀
Why this score: The speaker states the model is 'a lot more performant across all these different benchmarks' and that the 16GB model can run on 'most consumer devices' without showing any of the comparative benchmark data or defining what constitutes a 'consumer device' capable of running a 16GB AI model. While 16GB is relatively small for an 8B parameter model, 'most consumer devices' is a broad claim that might not hold true for many users without dedicated GPUs or sufficient RAM, as some top comments point out. The confidence in performance and accessibility isn't fully backed by presented evidence.
Original quote: “This means the same framework can be used for things like designing proteins or antibodies, predicting binding sites, or generating materials, and so on and so forth. Now, they released a family of three different models. The largest is 8 billion parameters, whereas the smallest is 1 billion. But…”
Logos models are 'a lot more performant' than competitors across benchmarks. — Confidence Mismatch (45/100)
At 10:30
Claims 'a lot more performant' but doesn't show the actual benchmarks or numbers. Just 'trust me, bro' vibes. 📈
Why this score: The speaker makes a strong claim about superior performance ('a lot more performant') but doesn't provide any visual evidence (like a chart or specific data points) to back it up. We're just told to 'see' it, which is a classic move to project confidence without showing the receipts.
Original quote: “Now, they released a family of three different models. The largest is 8 billion parameters, whereas the smallest is 1 billion. But as you can see, if you compare this with other similar competitors, this model is a lot more performant across all these different benchmarks, including ligand design…”
Introducing LTX Trainer as the official fine-tuning tool for specific AI creations. — No Frame (75/100)
At 12:30
Just laying out what the LTX Trainer does, seems pretty straightforward for an AI tool. No fancy footwork here.
Why this score: The speaker is describing the features and purpose of a specific AI tool, LTX Trainer, in a clear and factual manner. There's no apparent rhetorical manipulation or exaggeration; it's a direct explanation of functionality.
Original quote: “2 Trainer, which allows you to train and fine-tune this LTX model. For example, if you want a model to consistently generate a certain character or object or VFX, you can plug your data through this to train a LoRA from it. Now, this is a super flexible platform. It supports various workflows like…”
Pitches Codex's power for complex workflows, then immediately adds a catch 🚩 — Volume Game (45/100)
At 14:30
Hypes up Codex's power for 'hard to describe' workflows, then immediately whispers the 'doesn't work all the time' part. Classic volume game. 🤫
Why this score: The speaker first highlights the broad utility and power of Codex for any workflow, making it sound universally applicable. Immediately after, they introduce a significant limitation ('it doesn't work all of the time'), which downplays the initial grand claim. This is a volume game because the initial, louder claim is partially walked back by a quieter, less emphasized caveat.
Original quote: “You can see how powerful this is, especially for workflows where it's really hard to just describe with text. Instead, you can just take a screen recording of you doing the entire workflow. This could be like filing an expense or booking something, publishing a video, whatever your workflow might…”
The Ace robot 'absolutely dominated' a professional table tennis player. — Loaded Language (45/100)
At 16:30
Calling it 'absolutely dominated' without showing the full match or score is a bit much. We need receipts for that kind of smack talk. 🏓🔥
Why this score: The term 'absolutely dominated' is strong, emotionally charged language. While the robot might have won, 'dominated' implies a complete and overwhelming victory, which isn't explicitly shown or quantified in the clip. It's a subjective interpretation rather than a factual statement of the score or performance.
Original quote: “This one actually played against a professional human player and it absolutely dominated the game.”
Calling the Sony robot the 'best robot we have right now' for table tennis, based on its performance. — Confidence Mismatch (45/100)
At 18:30
Declaring it 'by far the best' is a bold claim, especially without showing other contenders or specific metrics. It's a vibe, not a benchmark. 🤖🏆
Why this score: The speaker is very confident in calling this robot 'by far the best' for table tennis. While the demo looks impressive, 'best' is a strong, subjective claim that would typically require comparative data against other top robots, which isn't provided. It's a confident assertion based on a single impressive demonstration, not a comprehensive review.
Original quote: “ball. This is much harder to train compared with just getting a paddle to hit the ball at any angle. Now this robot doesn't have legs so this is mounted on a heavy duty motorized rail, but it has incredible high speed latency and actuations. It has to like instantly move its entire mass from left…”
Introducing a new 'waifu' robot, Moya, with a face that 'doesn't really look human' 🤖 — No Frame (75/100)
At 20:30
Dude's just giving his honest take on the robot's face, even if it's a bit harsh. No tricks here, just vibes. 🤷♀️
Why this score: The speaker is describing the robot's features and offering a personal opinion on its realism. This is a subjective observation, not a factual claim being presented deceptively. It's a straightforward commentary on what's being shown.
Original quote: “teases their full body humanoid robot called Moya. You can see that this robot's face is moderately realistic. She can blink and tilt her head, and even demonstrate some facial expressions. And you can see how from this demo, this robot is intended for companionship. It's also able to do some…”
GLM 5.2 is the best open model released — Confidence Mismatch (45/100)
At 22:30
Calling it 'by far the best' is a bold claim, even if it's from their 'favorite lab.' Need more than just vibes. 🏆
Why this score: The speaker declares GLM 5.2 as 'by far the best open model' without immediately presenting the specific benchmark data to support such a strong, superlative claim. While they promise more details later, the initial statement is made with high confidence based on personal preference ('my favorite lab') rather than immediate, objective evidence.
Original quote: “This is an AI system that can help move”
GLM 5.2 is 'by far the best open model' — a bold claim! — Confidence Mismatch (45/100)
At 22:30
Calling GLM 5.2 'by far the best open model' is a big flex, especially without immediate data. Give us the receipts! 👑
Why this score: The speaker makes a strong, definitive statement ('by far the best open model') about GLM 5.2 without immediately providing the supporting evidence or benchmarks. While they promise more details later, the initial declaration is made with high confidence that isn't yet backed up.
Original quote: “below. Also this week, by far the best”
GLM 5.2 is the best frontier model for fields requiring factual accuracy like law or medicine. — Confidence Mismatch (45/100)
At 24:30
From 'lowest hallucination rate on THIS benchmark' to 'BEST frontier model for law/medicine' is a leap. That's a big jump in confidence. 🚀
Why this score: The speaker transitions from discussing a specific benchmark result (lowest hallucination rate on a tricky test) to a broad, definitive claim about GLM 5.2 being 'the best frontier model' for critical fields like law and medicine. While the benchmark is relevant, declaring it 'the best' for such high-stakes applications based solely on one type of test, without broader evidence or caveats, shows a confidence that might not be fully warranted by the presented data alone.
Original quote: “So, if you're working in a field that requires factually accurate information like law or medicine, GLM 5.2 is actually the best frontier model you can use right now.”
GLM 5.2 has the lowest hallucination rate according to 'Artificial Analysis omniscient's hallucination rate' benchmark. — Anonymous Authority (45/100)
At 24:30
Citing 'Artificial Analysis omniscient's hallucination rate' without a link or specific study name is a bit vague. Who's 'omniscient' here? 🧐
Why this score: The speaker references a specific benchmark ('Artificial Analysis omniscient's hallucination rate') but doesn't provide a direct source or link to this analysis. While the name sounds official, without a specific study, publication, or a way to verify it, it leans on an unnamed or vaguely defined authority. It's not necessarily false, but it lacks the transparency for full verification.
Original quote: “If you look at this Artificial Analysis omniscient's hallucination rate, you can see that by far GLM 5.2 has the lowest hallucination rate compared to the other frontier models.”
GLM 5.2 scored low due to lack of steerability. — Confidence Mismatch (45/100)
At 26:30
Went from 'seems like' to a definitive 'because' in one breath. That's a quick jump from speculation to fact. 🧐
Why this score: The speaker starts with 'it seems like' which suggests an inference or observation, but then immediately states 'because it's not as steerable' as a definitive reason. This implies a level of certainty that isn't fully supported by the initial hedging.
Original quote: “And it seems like it
scored so low because it's not as
steerable as the other frontier models.”
Claiming compressed models are still 'pretty usable' and not 'significantly dumber'. — Confidence Mismatch (45/100)
At 28:30
Went from specific stats to 'pretty usable' and 'not significantly dumber' — that's a vibe check, not a metric. What's 'pretty'?
Why this score: The speaker transitions from precise percentage points (76%, 82% accuracy) to subjective, qualitative assessments like 'pretty usable' and 'not significantly dumber'. This shift lacks the same level of evidence and precision, relying on a feeling rather than data.
Original quote: “So, it's still pretty usable. It's not like these compressed models are significantly dumber.”
Exaggerating the cost of running GLM 5.2, then offering a link. — Loaded Language (45/100)
At 28:30
Comparing running AI to selling a house or kidney is a bit much, even for a joke. Hyperbole for effect. 🏠💸
Why this score: The phrase 'sell your house or kidney' is a clear exaggeration used to emphasize the accessibility of the model. While likely intended as humor, it's loaded language designed to create a strong impression rather than convey literal fact.
Original quote: “You don't have to sell your house or kidney to run this. If you're interested in reading further, I'll link to this main page in the description below.”
Downplaying hardware requirements with a quick 'but you don't need that much' 🤏 — Volume Game (45/100)
At 30:30
Mentions a high-end GPU, then immediately waves it off. The 'but' is doing heavy lifting here. 🏋️
Why this score: The speaker quickly mentions a powerful H100 GPU, then immediately downplays its necessity without offering specific, accessible alternatives or benchmarks for 'mid-to-high-end consumer GPU' for the average user. It's a classic volume game: state a high bar, then quietly lower it without much detail, making it sound easier than it might be.
Original quote: “Here they say that they tested it on an H100 with 80 GB, but you don't need that much.”
Comparing new tech to MRI, but 100x faster — a bold speed claim — Confidence Mismatch (45/100)
At 32:30
Saying it's 'nearly 100 times the speed' of MRI without showing any actual data. That's a hell of a leap! 🚀
Why this score: The speaker makes a very specific, impressive claim about speed ('nearly 100 times the speed') by comparing it to MRI technology. However, no evidence or specific metrics are provided to back up this massive performance difference, making it a statement of high confidence without supporting data.
Original quote: “similar to today's MRI scanning”
The classic YouTube outro with a newsletter upsell. — Plain Sales Pitch (45/100)
At 34:30
Standard call to action, then a quick pivot to the 'exclusive' newsletter. Gotta get that email list, baby! 💸
Why this score: The speaker transitions from a general call to like/subscribe to a specific pitch for their 'free weekly newsletter' as the best way to 'really stay up-to-date.' It's a common tactic to leverage existing audience engagement into a direct marketing channel.
Original quote: “As always, I will be on the lookout for the top AI news and tools to share with you. So, if you enjoyed this video, remember to like, share, subscribe, and stay tuned for more content. Also, there's just so much happening in the world of AI every week. I can't possibly cover everything on my…”
See the full analysis with timestamps →