Last week we shipped Qwen3.8-2.4T-A95B on day zero. Today it becomes Qwen3.8-Max. ⚡️ https://do.co/4gA5TRY Same 2.4T parameter model, now with native image and video input and a 1M-token context window. If you're already calling it, same endpoint, nothing to change.
DigitalOcean
Software Development
Broomfield, Colorado 172,797 followers
AI-Native Cloud. ☁️
About us
DigitalOcean is the AI-Native Cloud purpose-built for the inference and agentic era. Its five-layer integrated platform—spanning GPU and CPU infrastructure, core cloud, inference, data, and managed agent orchestration—is open throughout with no vendor lock-in, giving builders everything they need to start fast, scale production AI workloads, and improve unit economics. More than 650,000 customers and millions of developers globally trust DigitalOcean to build, ship, and scale their applications.
- Website
-
https://www.digitalocean.com
External link for DigitalOcean
- Industry
- Software Development
- Company size
- 1,001-5,000 employees
- Headquarters
- Broomfield, Colorado
- Type
- Public Company
- Founded
- 2012
- Specialties
- Cloud Computing, Cloud Servers, Virtual Hosting, Cloud Hosting, Cloud Infrastructure, Simple Hosting, and Virtual Servers
Employees at DigitalOcean
View 2k employees at DigitalOcean
or
Locations
-
Primary
Get directions
105 Edgeview Dr
Broomfield, Colorado 80021, US
Updates
-
Close out Ray Summit right with Inferact, NVIDIA, Nous Research and DigitalOcean. 🍻 You'll hear about what Day 0 post-training for open models actually looks like, how to maximize your intelligence per dollar, what’s next for vLLM, and more. Mingle with peers and share what you’re excited to build next. RSVP today. 🔗 https://do.co/4g0oAhE
-
-
DigitalOcean Inference Router is now cache-aware. 🔀 https://do.co/3Uc5s7J Routing optimizes per request. Caching optimizes per session. Switch models mid-conversation and you lose the cache, along with the savings and consistency it bought you. Now Inference Router weighs the cost before switching and allows you to set a model affinity header or a routing budget. Built for coding assistants, long multi-turn agents, and any append-only workflow where the cache carries the weight.
-
Huge congratulations to Higgsfield AI on raising $400M in Series B financing! 🎉 It's been an honor to support Alex Mashrabov and the team on their journey to a $700M run rate, just 16 months after launch. Alex set out to give every storyteller the power that once took a Hollywood budget, and that vision is clearly resonating. One of AI's great growth stories, and you're just getting started. 🚀
-
-
DigitalOcean reposted this
We're back for Hacktoberfest 2026! This year, we're not counting PRs. Instead, we’re hosting 300+ in-person “Fests” and an online event all about building with open-source and open-weight AI. Learn more and see how you can get involved. https://hacktoberfest.com
-
DeepSeek AI-V4-Pro-0813 is now live on DigitalOcean Inference Engine. 🚀 https://do.co/4gA5TRY 1.6T parameters, 49B active, a 1M-token context window, and an Artificial Analysis Intelligence Index score of 53 at max reasoning effort, nearly double the open-weights median. Open weights, MIT license, fully managed infrastructure, usage-based pricing. Available via Serverless Inference and Inference Router.
-
To close out Ray Summit and the vLLM Conference, we're hosting a happy hour with NVIDIA and Inferact. Join us for lightning talks, live demos, cold drinks, and a room full of people who are serious about transitioning from tokenmaxxing to valuemaxxing. Space is limited, RSVP today. 🎟️ https://do.co/4g0oAhE
-
-
DigitalOcean reposted this
Chasing the leaderboard with one model is no longer enough. Last week I spoke at the 𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐀𝐈 𝐒𝐮𝐦𝐦𝐢𝐭 @UC Berkeley about why inference router is at the core of the AI strategy. One model for every task is overkill and creates unnecessary cost and risk. We built DigitalOcean’s 𝐈𝐧𝐟𝐞𝐫𝐞𝐧𝐜𝐞 𝐑𝐨𝐮𝐭𝐞𝐫 to change that. It’s a purpose-built, open-source layer that picks the right model for every request in under 200ms—cutting serving costs by up to 50% without vendor lock-in. Model orchestration is the new FinOps. It’s time to move past the leaderboard and start building for your team’s specific preferences. Watch this to know how we built the router. #AI #DigitalOcean #InferenceRouter #OpenSource #AgenticAI
-
DigitalOcean reposted this
Alibaba dropped Qwen 3.8 last week, and it's now on DigitalOcean Serverless Inference. The context window got me: hundreds of thousands of tokens in one pass. A whole codebase. So I built a little tool. You give it a GitHub repo, it reads the whole thing in one pass and tells you what it does, whether it's still maintained, and what it touches: network calls, secrets, shell, file writes. From the actual code, not the README. I install a ton of open-source and never know what most of it does. This reads a whole repo in a couple minutes, so I can check before I depend on something. It's a quick read to help you decide what to look at. It can miss things, so it's not a security audit, and it says so when it can't verify a language. Local, open-source tool. And yeah, I built an open-source tool to check open-source tools. The irony's not lost on me 🙂
-
Now available: Qwen3.8-2.4T-A95B, Alibaba Cloud’s most capable Qwen model yet, with 2.4T parameters, on DigitalOcean Serverless Inference via NVIDIA HGX™ B300 GPUs. 🤖 https://do.co/4gw3Ygj Built for long-horizon autonomous coding. Start building today. One API, usage-based pricing, and no infrastructure to manage.
