Futures
Access hundreds of perpetual contracts
TradFi
Gold
One platform for global traditional assets
Options
Hot
Trade European-style vanilla options
Unified Account
Maximize your capital efficiency
Demo Trading
Introduction to Futures Trading
Learn the basics of futures trading
Futures Events
Join events to earn rewards
Demo Trading
Use virtual funds to practice risk-free trading
Launch
CandyDrop
Collect candies to earn airdrops
Launchpool
Quick staking, earn potential new tokens
HODLer Airdrop
Hold GT and get massive airdrops for free
Launchpad
Be early to the next big token project
Alpha Points
Trade on-chain assets and earn airdrops
Futures Points
Earn futures points and claim airdrop rewards
More
Meta Releases SAM 3.1: 7x Speed Increase for 128 Target Tracking, Real-Time Video Segmentation
According to monitoring by 1M AI News, Meta’s Super Intelligence Lab has released the Segment Anything Model 3.1 (SAM 3.1), which is a direct upgrade from the visual segmentation foundational model SAM 3 released last November. The core improvement is Object Multiplex, a shared memory joint multi-target tracking method. Previously, SAM 3 required independent processing for each target, leading to a linear increase in inference costs with the number of targets. SAM 3.1 allows for the simultaneous processing of up to 16 targets in a single forward pass, achieving approximately 7 times inference acceleration when tracking 128 targets on a single H100 GPU, without sacrificing accuracy, by sharing global contextual information. Additionally, SAM 3.1 has shown performance improvements on 6 out of 7 video object segmentation (VOS) benchmarks. SAM 3.1 is a plug-and-play upgrade, and the model weights are available for download on Hugging Face.