Modal · Blogs

Blogs listings from Modal on TechiSeek. 155 public listings.

• Modal Clusters are generally available (Modal)
  Organizations need to own their intelligence to be successful. Training and serving that intelligence at scale, however, requires petaFLOP/…
• Run GPU jobs from Airflow with Modal (Modal)
  Isolate your tasks with Modal containers while using Airflow for orchestration.
• How Contextual AI automated CI with Modal GPUs (Modal)
  Learn how Contextual AI accelerated their developer iteration speed by using Modal to run tests on GPUs.
• Speculation Is All You Need (Modal)
  We are all-in on speculative decoding, and we’d like to tell you why. But first: we’re big fans of Z Lab’s DFlash draft model architecture.
• How sync. uses Modal to lipsync 100 hours of video a day (Modal)
  is a research lab training foundational models to understand and manipulate humans in video.
• How we used evals and inference-time compute scaling to generate beautiful QR codes that actually (Modal)
  Two techniques dominate discussion of the engineering of language model applications and artificial intelligence: evals and inference-time…
• Try GLM-5.1, the new frontier of open intelligence, on Modal | Modal Blog (Modal)
  GLM-5.1 establishes a new SotA for open models.
• Introducing: WebSockets on Modal (Modal)
  Modal now supports WebSocket connections, enabling real-time, bidirectional data transfer between client and server.
• Autoscaling Autoresearch: Give your agents elastic GPUs on Modal (Modal)
  Following Karpathy-san’s banger release of autoresearch, many are embracing the use of AI to accelerate AI research.
• Inside Modal Notebooks: How we built a cloud GPU notebook that boots in seconds (Modal)
  Exploring the internals of our new product, a modern Jupyter notebook built for fast startup and real-time collaboration.
• Introducing: B200s and H200s on Modal (Modal)
  We’re excited to be making Nvidia B200 and H200 GPUs available on Modal starting today!
• Kimi K3 by Moonshot now available on Modal (Modal)
  Kimi K3, a 2.8 trillion parameter multimodal model by Moonshot, along with a custom-trained DFlash speculator, is now available on Modal.
• Product updates: WebSocket support, interactive commands & more (Modal)
  We’ve been busy in 2024 so far, bringing you Web Sockets, interactive commands, H100 s and more.
• Competitive prompt engineering (Modal)
  Learn how Basis partnered with Modal to bring the spirit of competitive programming to prompt engineering.
• Transcribe speech 100x faster and 100x cheaper with open models (Modal)
  How we transcribed one week of audio in one minute for under one dollar.
• How to catch crypto miners using syscall signatures (Modal)
  See how Modal combats cryptomining abuse with syscall-based program analysis, to secure GPUs for legitimate users.
• Modal Sandboxes are generally available (Modal)
  Sandboxes are a new way to run code in Modal, with a focus on security and isolation.
• Scaling to 1 million concurrent sandboxes in seconds (Modal)
  How (and why) we built a scheduling system that can scale to 1 million concurrent sandboxes (per workspace) in seconds.
• Runtime Roundup: VM Sandboxes, Multi-node clusters, and more (Modal)
  Modal just hosted our inaugural conference, Runtime. Here are a few of the highlights that we announced.
• Twirl is joining Modal (Modal)
  Twirl, a Stockholm-based data orchestration platform, is joining Modal.
• We open sourced the GPU Glossary (Modal)
  GPU documentation for the people, now by the people.
• We reverse-engineered Flash Attention 4 (Modal)
  Asynchrony, fast approximate exponents, and 10x more efficient softmax.
• Inkling by Thinking Machines now available on Modal (Modal)
  Inkling, a general-purpose multimodal model by Thinking Machines, along with a custom trained DFlash speculator, is now available on Modal.
• Product updates: Price drops, brand refresh, and image-to-video example (Modal)
  Welcome to another round of Modal Product Updates!
• Product updates: Multi-node training clusters, B200 and H200s, and Client 1.0 release (Modal)
  Welcome to another round of Modal Product Updates!
• Making FlashAttention-4 faster for inference (Modal)
  When the Flash Attention-4 kernel source was released last year, we dove in and shared our findings about how the kernel works in excruciat…
• What is an AI code sandbox? (Modal)
  AI code sandboxes are seeing an explosion of adoption as the volume of LLM-generated code in the world grows.
• The future of AI needs more flexible GPU capacity (Modal)
  The last couple of years of Gen AI frenzy have brought us some undeniably cool new products, like Copilot and Suno.
• Real-time inference for robots at Physical Intelligence (Modal)
  How Physical Intelligence runs remote, real-time, robotic inference on Modal.
• Announcing our $87M Series B (Modal)
  We’re excited to announce that we have raised more than $80M in a Series B round, led by Lux Capital.
• Product Updates: Modal in AWS & GCP marketplaces, Sandbox improvements, and more (Modal)
  All the latest news across product, content, community, and events this month.
• What is LLM fine-tuning? (Modal)
  Fine-tuning helps us get more out of pretrained large language models (LLMs) by adjusting the model weights to better fit a specific task o…
• How to fine-tune an LLM on Modal (Modal)
  An operational guide to fine-tuning an LLM on any dataset in minutes (ft.
• How Decagon shipped real-time voice AI on Modal (Modal)
  How Decagon and Modal made real-time voice AI possible, combining fine-tuned small models with a re-engineered inference runtime for sub-se…
• Product updates: VM Sandboxes, Lower latency routing, RBAC, and more (Modal)
  Recent product updates and news from around the community.
• Modal is a computer (Modal)
  A machine that runs programs of arithmetic & logic operations on information.
• Devin Outposts on Modal (Modal)
  Devin, built by Cognition, is an AI software engineer: it plans, writes, tests, and ships code semi-autonomously.
• Quail: Speeding up AI-SQL by jointly optimizing query planner and inference engine (Modal)
  I see it as a point on the LLM pareto optimal curve in a regime that had a large revealed latent demand (no thinking, single token, low lat…
• Modal named to 2024 Enterprise Tech 30 list by Wing (Modal)
  Celebrating the best in enterprise tech.
• Justin Dignelli joins Modal as VP of Sales (Modal)
  Why I Joined Modal I first met Erik about three years ago over coffee, trading stories about the uphill battle of finding product-market fi…
• TypeSafe AI scales Jev to trillions of tokens in three days on Modal (Modal)
  TypeSafe AI launched Jev, a new class of AI model built for automation, and scaled to trillions of tokens over a weekend on Modal.
• Product updates: Sandbox Sidecars, new models, a refreshed dashboard, and more (Modal)
  Recent product updates from Modal and news from around the community.
• How Suno shaved 4 months off their launch timeline with Modal (Modal)
  Find out how Suno uses Modal to scale inference and batch pre-processing to thousands of GPUs.
• Scaling reinforcement learning at Applied Compute (Modal)
  How Applied Compute trains custom agents with Reinforcement Learning for enterprises like DoorDash, Cognition, and Mercor on Modal.
• Our first brand campaign (Modal)
  Behind the scenes of updating our visual identity and launching our first-ever out-of-home campaign in San Francisco.
• Achieve state-of-the-art inference latencies with speculative decoding (Modal)
  How Modal and Decagon worked together to cut inference latency - and you can too.
• Why Substack moved their AI and ML pipelines to Modal (Modal)
  Learn how Substack sped up their developer iteration cycles by moving ML training and deployment to Modal from AWS SageMaker.
• Bringing serverless functions closer to the speed of wire (Modal)
  Modal’s Function Call data path is now >50ms faster. Our new routing layer is geographically distributed, so you can further reduce your ne…
• Embedding English Wikipedia in under 15 minutes (Modal)
  Leverage Modal’s parallel batch jobs and in-house storage features to quickly generate embeddings for billions of tokens.
• One-second voice-to-voice latency with Modal, Pipecat, and open models (Modal)
  And we only relied on Modal, open weights models, and the open source framework Pipecat. Even if you’re not interested in voice AI per se b…