Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
UltraSane
17 days ago
|
parent
|
context
|
favorite
| on:
Gemini 3 Deep Think
ARC-AGI-3 uses dynamic games that LLMs must determine the rules and is MUCH harder. LLMs can also be ranked on how many steps they required.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
