Modal Labs GPU Pricing & Hourly Rates
Verified RatesTier 3 Datacenter (AWS/GCP/CoreWeave Colocation)Real-time verified rates (Updated 2026)
Modal Labs Serverless AI Token Pricing (4 Models)
| Model | Category | Cold Start | Context | Input ($/1M or Unit) | Output / 1M Tokens | Speed | Actions |
|---|---|---|---|---|---|---|---|
| Whisper-Large-v3 | AUDIO | WARM_CONTAINER (~1.0s) | — | $0.0030 / min | — | Stateless API | API Docs |
| ComfyUI-FLUX.1-dev | IMAGE | WARM_CONTAINER (~2.0s) | — | $0.020 / img | — | Stateless API | API Docs |
| DeepSeek-R1-Distill-70B | REASONING | WARM_CONTAINER (~1.5s) | 128K | $0.60 | $0.60 | 110 tps | API Docs |
| vLLM-Llama-3.3-70B | LLM | WARM_CONTAINER (~1.5s) | 128K | $0.65 | $0.65 | 120 tps | API Docs |
All Available Modal Labs GPU Instances (3 Tiers)
Click any GPU model to see live cross-cloud price comparisons against RunPod, TensorDock, Lambda Labs, and AWS.
| GPU Model & Specs | VRAM | Interconnect | vCPU / System RAM | Spot Hourly | On-Demand Hourly | Cost / GB VRAM | Actions |
|---|---|---|---|---|---|---|---|
| NVIDIA L40S 48GB | 48 GB | PCIe 4.0 x16 | 8 vCPUs / 32GB | $0.65/hr-19% | $0.80/hr | $0.017/hr | |
| NVIDIA A100 SXM4 80GB | 80 GB | NVLink (Fast) | 16 vCPUs / 64GB | $1.65/hr-11% | $1.85/hr | $0.023/hr | |
| NVIDIA H100 SXM5 80GB | 80 GB | NVLink (Fast) | 24 vCPUs / 128GB | $2.45/hr-11% | $2.75/hr | $0.034/hr |
Simulate active daily compute hours, stopped persistent disk fees, and compare total monthly cost on Modal Labs versus RunPod and AWS EC2.
Open Modal Labs Cost CalculatorCheck competing platforms like RunPod, TensorDock, and Lambda Labs to find lower storage fees or dedicated enterprise Tier 3 datacenters.
View Alternatives ComparisonModal Labs Pricing FAQ & Cost Guide
How do Modal Labs Spot vs On-Demand rates work?
On-demand instances give you uninterrupted compute until you voluntarily stop or terminate them. Spot instances offer dramatic discounts (typically 20% to 60% off), but may be interrupted or evicted if a higher-paying user bids on the host. Always enable automatic model checkpointing when using spot capacity.
Does Modal Labs charge for stopped instances or idle storage?
Yes. When an instance is stopped, GPU compute billing halts immediately, but attached persistent volumes continue to accrue storage fees ($0.15/GB/month Modal Volumes). To avoid unexpected charges, terminate instances or offload weights to S3/Cloudflare R2 when not in active use.
Are there hidden egress or bandwidth fees?
Egress on Modal Labs is billed at Free / Included within tier quota.
What payment methods does Modal Labs accept?
Modal Labs accepts standard credit card billing and enterprise invoicing with verified business accounts.