Modal
Modal Labs, Inc. · modal.com · Last checked:
- Bring-your-own-code training platform
- Serverless GPU training
- Inference deployment
Company status
Active, privately held.
Sources 1, 2, 3 · checked 2026-11-08
Capabilities
- Serverless GPUs billed per second (H100, A100 80GB, B200 and others) for running your own fine-tuning code
- Docs include an LLM training quickstart that trains Qwen3.5-4B on GSM8K with a reward function
- Run SFT, LoRA/QLoRA, full fine-tuning, or RL with open-source frameworks you choose
- Deploy trained models on the same platform
Sources 1, 2, 3 · checked 2026-11-08
Who it tends to fit
- Engineering teams that want full control of the training code and weights
- Startups wanting per-second GPU billing with no fixed cluster
Sources 1, 2, 3 · checked 2026-11-08
Pricing
GPU per second: H100 SXM5 $0.001097, A100 80 GB $0.000694, B200 $0.001736. Plans: Starter $0 + compute with $30/month free credits; Team $250/month + compute with $100/month free credits; Enterprise custom.
Sources 1, 2, 3 · checked 2026-11-08
Published certifications
Limitations to check
- No managed fine-tuning job API: you write and maintain training code, data pipelines, and evaluation
- HIPAA only on the Enterprise plan, and some products are out of scope for PHI
Sources 1, 2, 3 · checked 2026-11-08
Sources
- Per-second GPU prices, plan prices and credits, SOC 2 on all plans, HIPAA on Enterprisemodal.com/pricingChecked 2026-11-08
- SOC 2 Type II report in Trust Center; GDPR DPA; HIPAA on Enterprise plan with out-of-scope products listedmodal.com/docs/guide/securityChecked 2026-11-08
- LLM training quickstart example (Qwen3.5-4B, GSM8K, reward function)modal.com/docs/examples/unsloth_finetuneChecked 2026-11-08
Report a correction: [email protected]. Facts are re-checked at least every 90 days.