Sabitlenmiş Tweet
Announcing the 2026 ComputeSDK Scale Invitational.
The Maui Invitational for AI sandbox scaling. June 1st. RSVP Below.
100,000 concurrent sandboxes. One day. Every provider on the same playing field.
English
ComputeSDK
201 posts
ComputeSDK
@computesdk
The universal sandbox orchestrator. It's only one SDK, but any sandbox provider.
Katılım Şubat 2025
90 Takip Edilen1.2K Takipçiler
Early on in our benchmarking we discovered that speed and scale were most important to developers.
So let's see who can actually scale.
No marketing tweets. No talking heads. Just raw results shared publically. The good, the bad, and the ugy.
English
SCALE
So on June 1st @modal @tensorlake @RunloopAI @northflank @declaw_ai and @e2b have accepted the most ambitious sandbox scale test we've ever run.
100k sandboxes running concurrently.
English
🚨🚨🚨
We're back with Benchmark Friday, but in addition to our daily tests we've been working on the Scale Invitational.
The 100k Scale Invitational is the most ambitious sandbox benchmark.
See the thread🧵
English
Sandboxes: computesdk.com/benchmarks/san…
Storage: computesdk.com/benchmarks/sto…
Browsers: computesdk.com/benchmarks/bro…
English
On the browser front you can see the performance over time of @browserbase @usekernel @hyperbrowser is all neck and neck.
(we actually have a PR up to improve this benchmark itself.)
English
🚨🚨🚨
Benchmark Friday is back with results for storage, sandboxes, and browsers.
Looking for more detail views? See the links below!
English
Huge congrats to the good folks @latitudesh
Latitude.sh@latitudesh
We’re announcing long term contracts totaling $254M, bringing new customer commitments to $290M in the last 30 days. Complete offering across CPU, GPU, Network, and Storage, this is the power of combining @latitudesh with @megaportnetwork ⚡️ AI inference is moving to the edge. More regions. More deployments. More scale. More!
English
random fact: you don't need a full vm for forking
@TigrisData has support for it inside of their buckets.
GIF
Tigris Data@TigrisData
What if just-bash (from @vercel) had persistent storage? agent-shell mounts Tigris object storage as a filesystem: use real bash tools read/write like local files persist state across runs Same shell, now with durable state. 👇
English
Excited to have @e2b showing off their scale.
Every day we run a burst benchmark for 100 sandboxes.
On June 1st lets see what 100k sandboxes look like.
E2B@e2b
100K sandboxes concurrently? We're in for the challenge. The ComputeSDK team is benchmarking top providers under the same production-like load on June 1st. Same workload, same metrics, results published openly. Follow along to keep posted on the results of the benchmark event.
English
This past week we've met with so many of these providers.
Can't wait to see how the scale invitational develops in the coming weeks!
English
🚨🚨🚨
Benchmark Friday is back!
A huge shoutout to @tensorlake our latest provider added to the sandbox benchmark!
You can tell they've been working hard!
English
Excited to announce that that @declaw_ai is joining the scale invitational!
declaw_ai@declaw_ai
Same playing field, real workloads - real scale, public results. Let's go 🚀
English
🚨🚨🚨
Benchmark Friday is back and it's not just sandboxes.
We're benchmarking sandboxes, browsers, and storage now plus we'll be in SF next week.
English