๐Great to see @RedHat_AI and @poolsideai team up to make Laguna XS.2 faster and cheaper to serve in vLLM. A DFlash speculator built with Speculators drafts 8 tokens per forward pass for 2-3x faster decoding at no quality loss, and LLM Compressor adds FP8 / NVFP4 / INT4 checkpoints so you can match your hardware budget.
๐ https://t.co/3mguMKtwTt
๐ Command A+ is ready to serve on vLLM: day-0.
Frontier open-source, production-ready. Huge congrats to the Cohere and vLLM teams!
Read more ๐
https://t.co/VWAeYGDmAR
Arc Hackathon Spotlight: Chariot
Chariot is an Arc Hackathon Spotlight project exploring how lending infrastructure can work across chains with ETH-backed collateral, vault logic, and automated risk controls.
Join @samconnerone and Sipho Yawe live on May 7 at 11:00 AM ET for a walkthrough of the build:
โ ETHEscrow to BridgedETH
โ ERC-4626 vault design
โ Liquidation logic and circuit breakers
โ Vault agent architecture
See how hackathon projects on Arc turn into real, composable financial infrastructure.
RSVP: https://t.co/lYDm8wnhWR
๐๐๐ https://t.co/SibONtoSCo just got merged to main!
Huge shoutout to the entire team @inferact that worked on day-0 support of DeepSeek V4 and our partner @NVIDIAAI for the collaboration on day-0 large scale serving enablement!
More optimizations coming soon - stay tuned!
If you're offering compute but egress is eating up your margins, let's talk
If your training data is stuck in one region and your GPUs are in another, let's talk.
If your content is being consumed, but you don't know by whom, let's talk.
Booth E11. We're here all day.
@Crestlayer has officially become a validator node on the Canton mainnet.
@CantonNetwork is the first privacy-enabled Layer 1 blockchain designed specifically for institutional finance, enabling regulated entities to securely, compliantly, and interoperably transfer assets.
Today, cross-border payroll routes through various intermediaries, each taking a fee.
With stablecoins, payroll providers get lower-cost payouts, faster account funding, and new revenue streams.
A practical guide for payroll platforms:
https://t.co/x5TO9SOzlc
@rpranav Decentralized systems avoid this by aligning incentives over the long term, which is essential for durable infrastructure.
As a professional infrastructure operator, we would like to ask whether there is an opportunity to contribute during the early testnet phase.
The stakes have changed in
@CantonNetwork
: With this halvening the majority of $CC rewards goes to apps and users - more than 60% of all CC minted! TradFi and crypto firms alike are competing for the same pool of rewards and everyone wins as adoption grows.
@rpranav Well put.
Without predictable low-latency access, most of these use cases simply canโt function โ decentralization has to come with performance.
Impressed by what @BitRobotNetwork is building โ AI automation agents that aim for dependability and real execution, not sandbox experiments.
This focus on reliable orchestration and execution is exactly what the future of autonomous systems needs. Looking forward to updates!
@rpranav@shelbyserves The angle on decentralized, low-latency storage for data-heavy apps and AI agents really resonates. Looking forward to seeing Shelbyโs testnet progress.
Really enjoyed this conversation โ especially the breakdown on hot storage vs cold storage and the latency tradeoffs vs Filecoin / Arweave.
@rpranav@shelbyserves
โ๏ธ Deep dive into decentralized hot storage with @rpranav (Head of Core Infra @AptosLabs & Lead @shelbyserves)
I asked him why Web3 needs "Amazon S3 but decentralized" and how they plan to beat cloud giants on speed and price.
Timestamps:
๐ฆ 00:00 โ What is @shelbyserves
๐ธ 00:53 โ The "Cloud Bill" problem: Why Web2 storage hurts startups
๐๏ธ 02:10 โ Shelby vs Filecoin/Arweave: Milliseconds vs Seconds
๐๏ธ 02:59 โ The 6 technical pillars to achieve speed and low latency
๐ 05:23 โ Privacy: Token-gated data & on-chain access control
๐ฌ 06:28 โ Use cases: Decentralized YouTube/TikTok & Prediction Markets
๐ค 07:16 โ The fastest growing sector: Storage for AI Agents
๐ 07:40 โ Why Shelby is 4x cheaper
๐๏ธ 08:55 โ Roadmap: Wen Testnet & Road to Mainnet
๐ช 10:58 โ Economics: Payments, token
๐ 11:35 โ Why Aptos is the only chain that can handle Shelby's load
๐ก๏ธ 12:26 โ Pranavโs personal investment strategy ("I'm all in ...")
If you are building data-heavy apps โ this conversation is crucial.
Very bullish on the direction @ShelbyServes is taking.
Real-time, decentralized data access is something the ecosystem has been missing for a long time.
Liquidity fragmentation rarely breaks markets outright. It slowly increases the cost of being wrong. You notice it when spreads widen exactly when you need them tight, when liquidations arrive sooner than models suggested. What gets labeled as volatility is often inefficiency