Luis Capelo retweetledi
Day 0 support for Step 3.7 Flash on Modal.
- 198B parameter MoE with 11B active
- 256K context
- 3 reasoning levels
- Native image & video understanding
Great to work with @StepFun_ai and @sgl_project on this one.
English
Luis Capelo
1.3K posts
Luis Capelo
@luiscape
Interested in deep learning research and applied AI systems. Building serverless at @modal
New York, NY Katılım Mayıs 2009
3.4K Takip Edilen1.6K Takipçiler
Luis Capelo retweetledi
Luis Capelo retweetledi
Modal is proud to now support @claudeai Managed Agents with Modal Sandboxes.
Claude@claudeai
Live from Code with Claude London: we're launching self-hosted sandboxes (public beta) and MCP tunnels (research preview) in Claude Managed Agents. Run agents inside your own perimeter, with your security controls applied by default.
English
Luis Capelo retweetledi
Step 4 to achieve truly serverless GPUs for AI inference: skip over unserializable inference engine setup steps like CUDA graph capture and Torch compilation by stacking GPU snapshots and CPU snapshots.
English
English
@botir33751732 @modal Happy to chat with someone from @modal to see how this can work with our usecase. Can someone tag anyone I can talk to below?
English
Luis Capelo retweetledi
Luis Capelo retweetledi
The @runwayml API Hackathon.
May 8 to 11.
Powered by Modal.
Runway@runwayml
Our first online Runway API Hackathon begins next week. One weekend. Free API credits. $25,000 in prizes. Build real-time video agents, agentic media pipelines or anything you can imagine. No experience required. Learn more and register at the link below.
English
Luis Capelo retweetledi
250 tok/s decode on a single-user setup in production is seriously strong 🚀 Nice SGLang! Go Modal!
Charles 🎉 Frye@charles_irl
Try it today on @modal! For coding tasks with high specdec, I saw up to 250 decode tok/s from @sgl_project for a single user of V4 Pro on an 8xB200. modal.com/docs/examples/…
English
@lemos_ronaldo Vale muito a pena data centers no Brasil mas como faz pra avançar a questão do imposto de importação? O primeiro passo é a remoção dos impostos nas GPUs da NVIDIA e todos os materiais relacionados a data centers.
Português
🚨 Falei hoje no Senado sobre data centers.
🟢 O Brasil tem hoje a combinação mais favorável do planeta para liderar IA sustentável: grid 10x mais limpo que EUA, maior potencial eólico-solar, cabos submarinos em Fortaleza, e biometano com potencial de 120 milhões m³/dia. Segue🧵
Português
Luis Capelo retweetledi
One portal, unlimited possibilities.
You can now access Modal via Tool Gateway by @NousResearch, makers of Hermes Agent. Check it out 👇
Nous Research@NousResearch
Tool Gateway is now live in Nous Portal. No separate accounts, no API key juggling. All you need is one subscription, and everything works. A paid Nous Portal subscription now includes access to 300+ models and a growing set of third-party tools. Launching with: → Web scraping → Browser automation → Image generation → Cloud terminal backend → Text-to-speech
English
Luis Capelo retweetledi
To show off what you can do with @OpenAI Agent SDK + @modal, we built an ML research agent (inspired by @karpathy).
It can:
- Spin up GPU sandboxes of any shape
- Run a pool of subagents
- Persist memory
- Snapshot state for fork/resume
Here it is playing Parameter Golf:
Modal@modal
Agents need computers. And they need a lot of them. Modal is an official sandbox provider for the @OpenAI Agents SDK.
English
Luis Capelo retweetledi
So happy the acquisition is public so I can finally post about the cool shit we’re building at @modal
This was a fun project:
Akshat Bubna@akshat_b
To show off what you can do with @OpenAI Agent SDK + @modal, we built an ML research agent (inspired by @karpathy). It can: - Spin up GPU sandboxes of any shape - Run a pool of subagents - Persist memory - Snapshot state for fork/resume Here it is playing Parameter Golf:
English
@BuiltByJulian @modal Hey @BuiltByJulian, we're looking into it. Are you in our Slack by any chance? We can chat and figure out what happened. Sorry for this!
English
To every builder: Don't use @modal to run AI models. I set it up some weeks ago and use it in production across 3 different apps. Since yesterday all calls are failing and I can't login, I assume they suspended my account without notifying me, their support is unresponsive.
English
Luis Capelo retweetledi
Build long-running agents with more control over agent execution.
New capabilities in the Agents SDK:
• Run agents in controlled sandboxes
• Inspect and customize the open-source harness
• Control when memories are created and where they’re stored
English
Luis Capelo retweetledi
Luis Capelo retweetledi
The future of artificial intelligence is physical.
@physical_int runs robotic control inference on Modal with >2x lower latency than the lag between your brain and your finger.
English
Luis Capelo retweetledi
We partnered with @Zai_org to bring GLM-5.1 to Modal. Free to try as an endpoint for the next month.
GLM-5.1 further improves upon GLM-5's coding abilities and long-horizon effectiveness.
Z.ai@Zai_org
Introducing GLM-5.1: The Next Level of Open Source - Top-Tier Performance: #1 in open source and #3 globally across SWE-Bench Pro, Terminal-Bench, and NL2Repo. - Built for Long-Horizon Tasks: Runs autonomously for 8 hours, refining strategies through thousands of iterations. Blog: z.ai/blog/glm-5.1 Weights: huggingface.co/zai-org/GLM-5.1 API: docs.z.ai/guides/llm/glm… Coding Plan: z.ai/subscribe Coming to chat.z.ai in the next few days.
English
Luis Capelo retweetledi
we've been cooking :)
Zhijian Liu@zhijianliu_
DFlash just landed in both SGLang and vLLM! 🚀 More draft models dropping soon: GLM-5.1, Kimi-K2.5 (preview live now!), Qwen3.5-397B & 122B. Try it now ↓ SGLang: github.com/sgl-project/sg… (🙏 @_dcw02) vLLM: github.com/vllm-project/v… (🙏 @BenjaminCh44989)
English
Luis Capelo retweetledi
New in the @modal docs: deploy an H200 server for Gemma 4 26B A4B with @vllm_project.
Not optimized for performance, but still gets ~175 tok/s at concurrency 1.
modal.com/docs/examples/…
English
Luis Capelo retweetledi
mngr: programmatically manage 100s of claude code sessions in parallel 🤖
open source today.
lets you do things like:
— for each open GitHub issue, create a PR
— for each flaky test in the past week, fix it
— for each rule in style guide, scan codebase & fix all instances
runs any agent: @claudeai, codex, @opencode, etc.
runs on any compute: locally, @modal, @Docker, or anything you can ssh into.
GIF
English