To win trading on Solana at scale, you need shreds from @SolanaCDN
Choose wisely
you met me at a very green time of my life
The name's Shreds.
Fastest Shreds.
Hear me out
Solana validator in your pocket 👀
heating up
Winning on @solana, coming right up.
Trading will never be the same again on Solana
This is an announcement of an announcement
We spent 16 hours testing an app on Solana from an unstaked sender. The results surprised us.
The good: median tx inclusion was under 400ms, with no stake, Jito, priority fees, or stake-weighted QoS. p99 was ~2.8s. 597 of 647 validators accepted 90%+ of what we sent.
The bad: ~6.5% of transactions simply disappeared. And geography mattered a lot: same-slot inclusion was 78% when co-located vs. 56% from home.
The takeaway: Solana really is fast and deterministic. But for unstaked users, stake and geography still define the edge.
Show more
Most reliable acceleration layer for @solana, check
AI inference layer, check
[REDACTED], check mate
Muse Glimmer quants on the way 🫡
Congrats to the @AIatMeta team on the release.
Today we're also opening the weights for Muse Glimmer, a great 30B parameter dense model that can run locally. Soon we'll also release the weights for Muse Spark 1.2, our latest foundation model. Meta is a strong supporter of open source and I'm proud of these releases. Congrats to @alexandr_wang and the MSL team for all your great work on these models.
Show more
Which Qwen 3.8 quant do you want?
1 model and a dream.
That's all you need.
Run Kimi K3 on a Mac Studio 🫰
K3 is 2.8T parameters and 1.6TB on disk, which makes it impossible to run on Apple Silicon.
Until now.
Our MLX port is now open source:
To accomplish this, we solved two things:
1. We wrote a streaming converter that walks one layer at a time, so that mlx_lm doesn't need to materialize the whole model.
2. REAP pruning sits on top and scores all 896 experts against a calibration corpus to keep only ones your workload needs.
That's what brings K3 down to 350GB and inside a Mac Studio.
Show more
