Gave four models the same text description of elon and asked them to draw him.
Qwen3.8 max vs fable5 vs gpt5.6 sol vs kimi k3. some got the face scary close, others made a whole new guy. wild how different the reads are.
Ran it all in https://t.co/tw7phbA3ir.
https://t.co/tw7phbA3ir turned a 6-year-old's robot question into a Lego movie starring Elon. Fixed Musk minifig, lab, circuits to bipedal walk �� uploaded "Optimus Tech" and got a full bot journey.
The kid thinks Elon is actually Lego now.
#LegoOptimus #AIVideo #Medeo
@elonmusk I benchmarked 4 LLMs and 2 video models on one concept.
Seedance 2.0 won raw video quality. Kimi K3 won on cost. But Grok’s full-stack pipeline scored highest overall (90.6) by minimizing handoff drift.
Fewer handoffs. More fidelity.
Full results ↓
A fixed Lego Elon, a futuristic garage, and a journey from battery science to self‑driving tech. Upload "Tesla Tech" — AI builds the mission.
#MedeoAI#LegoTesla