The Curious Engineer

All blogs / Distributed systems / cloud

Modal Blog

modal.com/blog

We share some insights about serverless computing, and the problems that we solve along the way.

Distributed systems / cloudAI / MLDevOps / SRE / infrastructure

’23’24’25’26
Posts
153 from 15 Jun 2023 to 23 Oct 2026
Posting
weekly (71 in the last year)
Kind
company blog
Feed
https://modal.com/blog/atom.xml

2026 63

  1. Launching a trillion-parameter model at Sakana AI
  2. Introducing our Training API
  3. Get started with LLM training on Modal
  4. Serverless routing made sticky
  5. NVIDIA Vera CPUs double throughput for Modal Sandboxes
  6. How Legora runs 2,000 parallel agent evals on Modal Sandboxes
  7. Building fast filesystems for VMs
  8. Augment Your LiveKit Agent with Modal GPUs
  9. Muse Spark Open is now available on Modal
  10. Modal crosses $1B in Annualized Run Rate and raises $750M Series D
  11. Toward RL on mixed-precision, mixed-accelerator samplers
  12. Endpoint Candidates: evaluate and promote inference endpoints
  13. TypeSafe AI scales Jev to trillions of tokens in three days on Modal
  14. Modal Clusters are generally available
  15. VM Sandboxes: Full computers for agents
  16. Runtime Roundup: VM Sandboxes, Multi-node clusters, and more
  17. Sidecars: A low-latency trust boundary for Sandboxes
  18. Quail: Speeding up AI-SQL by jointly optimizing query planner and inference engine
  19. How to serve trillions of tokens for trillion-parameter coding agents
  20. Product updates: Sandbox Sidecars, new models, a refreshed dashboard, and more
  21. One compute platform for end-to-end training at Markov Robotics
  22. Modal is expanding in Europe with our new London office
  23. How Botika runs full-stack generative AI on Modal
  24. Qwen3.8-2.4T-A95B now available on Modal
  25. Bringing serverless functions closer to the speed of wire
  26. A note on the Hugging Face agent incident
  27. Kimi K3 by Moonshot now available on Modal
  28. Devin Outposts on Modal
  29. Scaling to 1 million concurrent sandboxes in seconds
  30. Inkling by Thinking Machines now available on Modal
  31. Modal is a computer
  32. How to price serverless GPUs
  33. Multi-token Residual Prediction
  34. Anthropic integration with Modal brings scalable compute to Claude Science
  35. Modal's serverless Servers
  36. Achieve state-of-the-art inference latencies with speculative decoding
  37. Introducing Modal Auto Endpoints: Optimized inference you actually own
  38. Unpacking sandbox startup latency: why started ≠ ready
  39. Speculation Is All You Need
  40. Product updates: VM Sandboxes, Lower latency routing, RBAC, and more
  41. Making FlashAttention-4 faster for inference
  42. Reinforcement learning is an infrastructure problem
  43. Role-Based Access Control for humans and agents
  44. Modal's Series C: Raising $355M at a $4.65B valuation
  45. Scaling reinforcement learning at Applied Compute
  46. Introducing Claude Managed Agents with Modal Sandboxes
  47. How we achieved truly serverless GPUs
  48. Boosting multimodal inference performance by >10% with a single Python dictionary
  49. Building an RL theorem-proving workflow on Modal
  50. Building with Modal and the OpenAI Agents SDK
  51. Autoscaling Autoresearch: Give your agents elastic GPUs on Modal
  52. Butter is joining Modal
  53. Real-time inference for robots at Physical Intelligence
  54. Product Updates: RTX Pro 6000 Blackwell, Command K, Sandbox FS API and more
  55. Runway chooses Modal to power real-time inference for Runway Characters
  56. How Doppel eliminated ML infrastructure tax with Modal
  57. Product Updates: Directory Snapshots, GLM-5, billing updates and more
  58. Accelerating AI research that accelerates AI research
  59. Directory Snapshots: Resumable project state for Sandboxes
  60. How Ramp built a full context background coding agent on Modal
  61. Try GLM-5.1, the new frontier of open intelligence, on Modal
  62. Product Updates: Modal in AWS & GCP marketplaces, Sandbox improvements, and more
  63. Seamless computational bio at Chai Discovery

2025 47

  1. Keeping 20,000 GPUs healthy
  2. Modal + Mistral 3: 10x faster cold starts with GPU snapshotting
  3. Agents need good developer experience too
  4. How Reducto improved enterprise-scale document processing latency by 3x
  5. Host overhead is killing your inference efficiency
  6. How Decagon shipped real-time voice AI on Modal
  7. One-second voice-to-voice latency with Modal, Pipecat, and open models
  8. Product updates: Updates to Volumes, JS and Go SDKs, and more
  9. Modal + Datalab: Deploy high-throughput document intelligence in <5 minutes
  10. Announcing our $87M Series B
  11. We reverse-engineered Flash Attention 4
  12. Justin Dignelli joins Modal as VP of Sales
  13. Build an AI coding platform that scales to millions of monthly sessions
  14. Product updates: Modal Notebooks, sandbox idle timeouts, and more
  15. Inside Modal Notebooks: How we built a cloud GPU notebook that boots in seconds
  16. Introducing Notebooks
  17. How Zencastr transcribed hundreds of years worth of audio in just a few days
  18. Product updates: GPU memory snapshots, notebooks, service tokens, and more
  19. GPU Memory Snapshots: Supercharging sub-second startup
  20. What is an AI code sandbox?
  21. Transcribe speech 100x faster and 100x cheaper with open models
  22. Dollars per token considered harmful
  23. Product updates: Multi-node training clusters, B200 and H200s, and Client 1.0 release
  24. Jamsocket is joining Modal
  25. How Modal powered 250,000 Lovable app creations in a weekend
  26. How we used evals and inference-time compute scaling to generate beautiful QR codes that actually work
  27. How Quora uses Modal to run thousands of Python sandboxes simultaneously
  28. Run FLUX.1-dev three times faster
  29. Introducing: Modal 1.0
  30. Introducing: B200s and H200s on Modal
  31. Twirl is joining Modal
  32. Introducing Modal Batch: Process 1 million jobs with 1 line of code
  33. Modal's serverless KV store gets its limit raised to infinity
  34. Linear programming for fun and profit
  35. Modal SDKs for JavaScript and Go (alpha)
  36. How Lemon Slice built real-time generative video with Modal and Daily
  37. How sync. uses Modal to lipsync 100 hours of video a day
  38. Product updates: Running batch jobs with 1M inputs, ephemeral apps, and a new TensorRT-LLM example
  39. Our first brand campaign
  40. Product updates: Price drops, brand refresh, and image-to-video example
  41. 'I paid for the whole GPU, I am going to use the whole GPU': A high-level guide to GPU utilization
  42. We open sourced the GPU Glossary
  43. Product updates: Logs v2, live container profiling, and region selection on all plans
  44. Memory snapshots: Checkpoint/restore for sub-second startup
  45. Product updates: L40Ss, proxy auth tokens, and sandbox disk snapshotting
  46. Modal Sandboxes are generally available
  47. Modal is SOC 2 Type II compliant

2024 39

  1. Product updates: memory snapshotting, OIDC, async job queues & more
  2. Introducing: L40S GPUs on Modal
  3. What is LLM fine-tuning?
  4. Dogfooding Modal: What we learned at our internal hackathon
  5. How a top tier European soccer team sped up their data processing and reduced costs by 50%
  6. WireGuard at Modal: Static IPs for serverless containers
  7. Press release: Modal signs strategic collaboration agreement with AWS to deliver accelerated generative AI solutions
  8. How OpenArt scaled their Gen AI art platform on hundreds of GPUs
  9. Product updates: Static IP proxies, Slack integration, live usage dashboard & more
  10. Tidbyt is joining Modal
  11. The future of AI needs more flexible GPU capacity
  12. Hybrid search over California embeddings with Modal, MongoDB, and Clay
  13. How Contextual AI automated CI with Modal GPUs
  14. Boost your throughput with dynamic batching
  15. Building a cost-effective analytics stack with Modal, dlt, and dbt
  16. Product updates: Rollbacks, batching, sandbox tunnels & more
  17. Modal supports HIPAA compliance
  18. Inside the Modal code playground
  19. GPU prices are falling...
  20. Beat GPT-4o at Python by searching with 100 dumb LLaMAs
  21. Product updates: Datadog integration, lower function latency & more
  22. Competitive prompt engineering
  23. Run GPU jobs from Airflow with Modal
  24. How to catch crypto miners using syscall signatures
  25. How Hunch supercharged AI workflows with Modal Sandboxes
  26. Create an infinite icon library by fine-tuning Stable Diffusion
  27. Why Substack moved their AI and ML pipelines to Modal
  28. Introducing: Region selection
  29. Product updates: Cloud buckets, Okta SSO & more
  30. Beating proprietary models with a quick fine-tune
  31. Why you should move your ETL stack to Modal
  32. Modal named to 2024 Enterprise Tech 30 list by Wing
  33. How Ramp automated receipt processing with fine-tuned LLMs
  34. Lambda on hard mode: Inside Modal's web infrastructure
  35. Introducing: WebSockets on Modal
  36. How Suno shaved 4 months off their launch timeline with Modal
  37. Product updates: WebSocket support, interactive commands & more
  38. Introducing: H100s on Modal
  39. Embedding English Wikipedia in under 15 minutes

2023 4

  1. How to fine-tune an LLM on Modal
  2. Modal is now generally available
  3. Press release: Modal Labs announces Series A financing round
  4. Modal is SOC2 compliant