[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

# ![@CarIntel Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::1982910763217166336.png) @CarIntel Wild Child

Wild Child posts on X about ai, xpeng, tesla, 6969 the most. They currently have XXX followers and XX posts still getting attention that total XX engagements in the last XX hours.

### Engagements: XX [#](/creator/twitter::1982910763217166336/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1982910763217166336/c:line/m:interactions.svg)

- X Week XXX +440%
- X Month XXXXX +70%

### Mentions: X [#](/creator/twitter::1982910763217166336/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1982910763217166336/c:line/m:posts_active.svg)


### Followers: XXX [#](/creator/twitter::1982910763217166336/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1982910763217166336/c:line/m:followers.svg)

- X Week XXX +30%
- X Month XXX +281%

### CreatorRank: undefined [#](/creator/twitter::1982910763217166336/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1982910763217166336/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[stocks](/list/stocks)  [automotive brands](/list/automotive-brands)  [technology brands](/list/technology-brands) 

**Social topic influence**
[ai](/topic/ai), [xpeng](/topic/xpeng) #102, [tesla](/topic/tesla) #2446, [6969](/topic/6969) #1261, [math](/topic/math), [intuition](/topic/intuition), [li auto](/topic/li-auto), [the first](/topic/the-first), [future](/topic/future), [bmw](/topic/bmw)

**Top accounts mentioned or mentioned by**
[@grok](/creator/undefined) [@rryssf](/creator/undefined) [@noahpinion](/creator/undefined) [@proffeynman](/creator/undefined) [@stijnnoorman](/creator/undefined) [@iyoushetwt](/creator/undefined) [@cgtwts](/creator/undefined) [@brianarmstrong](/creator/undefined) [@nntaleb](/creator/undefined) [@sophiacycles](/creator/undefined) [@cremieuxrecueil](/creator/undefined) [@bjornlomborg](/creator/undefined) [@blackhackofdoom](/creator/undefined) [@mrewanmorrison](/creator/undefined) [@abhinavxj](/creator/undefined) [@helloiamleonie](/creator/undefined) [@waymo](/creator/undefined) [@tesla](/creator/undefined) [@xpengmotors](/creator/undefined)

**Top assets mentioned**
[XPeng Inc. (XPEV)](/topic/xpeng) [Tesla, Inc. (TSLA)](/topic/tesla) [Li Auto Inc. (LI)](/topic/li-auto)
### Top Social Posts
Top posts by engagements in the last XX hours

"Difference between E2E Adas model and VLM. Before E2E self-driving cars had separate AI models for each task like a relay race: X. Detector AI: "I see a car." X. Tracker AI: "The car is moving left." X. Predictor AI: "It will likely cut us off." X. Planner AI: "I will brake." Nvidia paper UniAD's breakthrough was melting these four separate AIs into one single Transformer. It proved that merging it all car drives better. The "thoughts" inside UniAD are pure math (vectors) not words"  
[X Link](https://x.com/CarIntel/status/1997739803089961243)  2025-12-07T18:47Z XXX followers, XX engagements


"VLM think words. Li uses VLM as second logical brain. Something similarly Xpeng is doing. Xpeng focuses on using VLAs with their xBrain Architecture. Xpeng relies aggressively on distilling the cloud brain. Now they are neck2neck race. Gf"  
[X Link](https://x.com/CarIntel/status/1997741425136652334)  2025-12-07T18:54Z XXX followers, XX engagements


"E2E Learning (FSD12) threw away the 300000 lines of human code. CNNs are still the eyes but now a single giant AI model processes it (using "Attention" to understand context just like ChatGPT). The car isn't following rules: it is imitating human intuition. If it sees a man in a chicken costume it thinks "I've seen humans slow down when weird stuff is on the road so I will too.""  
[X Link](https://x.com/CarIntel/status/1997736341593809193)  2025-12-07T18:33Z XXX followers, XX engagements


"It is the only mass-produced E2E system you can buy and use on the road today. However Wayve was the first one to prove it. Chinese players go even a step further: Li Auto has Kahneman System X / System X approach. System X is E2E: fast instinctual driving like staying in a lane. System X (VLM): thinks slowly about complex situations (like seeing a construction sign and reading the text)"  
[X Link](https://x.com/CarIntel/status/1997736345481920517)  2025-12-07T18:33Z XXX followers, XX engagements


"Cheat sheet for autonomous driving AI: X. Classic: CNNs + SW + Action (BMW L3) X. End-to-End (E2E) (Tesla FSD): Exchange SW with transformer XXX. E2E + VLM (Li with Lingo1 and Lingo2 Wayve) X. VLA (XPeng/Wayve). I reason I speak X. Generative World Model (Sora). "I imagine the movie of the future." (Visual Prediction). X. JEPA (LeCun). "I calculate the physics of the future." (Abstract Prediction)"  
[X Link](https://x.com/CarIntel/status/1997787761894445368)  2025-12-07T21:58Z XXX followers, XX engagements


"XPENG delivered cars with insane amount of compute power: 2250 TOPS (roughly 4x-5x more than a Tesla HW4). At first no one understood why. Until they announced VLA 2.0: "VisionImplicit TokenAction" Old Way (VLM): You compress the rich video into a simple word: "Red Light." XPeng (Implicit): You keep the rich high-dimensional math that represents the "Concept of Red Light" without simplifying it into a word. It is likely that XPeng is not just running one VLA. To reach true L4 safety they might be running two in parallel (or a large VLA + a World Model verifier)"  
[X Link](https://x.com/CarIntel/status/1997801943100731872)  2025-12-07T22:54Z XXX followers, XX engagements


"Most aggressive L4 companies in my eyes: XPENG and Tesla. XPeng embraces Yann LeCuns vision who critiques pixel/word prediction as inadequate for real-world tasks favoring world models that internalize physics causality and dynamics. XPengs physical AI implements this in VLA XXX (november 2025)"  
[X Link](https://x.com/CarIntel/status/1997929643349815576)  2025-12-08T07:22Z XXX followers, XX engagements


"That is why in all prompt engineering courses you learn to start your prompt with imagine you are in the role of"  
[X Link](https://x.com/CarIntel/status/1997949417215922558)  2025-12-08T08:40Z XXX followers, XX engagements


"The story behind Stable Diffusion algorithm started with a physicist Jascha Sohl-Dickstein in 2015 and a paper Deep Unsupervised Learning using Nonequilibrium Thermodynamics. He looked at how a drop of ink spreads (diffuses) into a glass of water until it is just a cloudy mess. He asked If we can mathematically describe how the ink destroys the clear water can we reverse the math to turn the cloudy mess back into a clear drop of ink"  
[X Link](https://x.com/CarIntel/status/1997724081462862074)  2025-12-07T17:45Z XXX followers, XX engagements


"Then the race began: OpenAI Sora under the hood was a Diffusion Transformer (02/24). In 06/24 Stability AI released SD3 which had Transformer and Inventors of original Stable Diffusion (Rombach) made a company Black Forest Labs and released Flux X. Flux is currently considered the king"  
[X Link](https://x.com/CarIntel/status/1997726418784968732)  2025-12-07T17:54Z XXX followers, XX engagements


"Who will win @Waymo or @Tesla Or @XPengMotors I bet XPENG anf Tesla at the same time"  
[X Link](https://x.com/CarIntel/status/1998469695796609433)  2025-12-09T19:07Z XXX followers, XX engagements