lambda-labs-gpu-cloud
Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent filesystems, or high-performance multi-node clusters for large-scale training.
Install this skill
Works with Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, and 70+ other harnesses.
npx skills add KalarisLabs/research-agent-skills --skill lambda-labs-gpu-clouduvx research-agent-skills install lambda-labs-gpu-cloudDocumentation & Instructions
Lambda Labs GPU Cloud
Comprehensive guide to running ML workloads on Lambda Labs GPU cloud with on-demand instances and 1-Click Clusters.
When to use Lambda Labs
Use Lambda Labs when:
- Need dedicated GPU instances with full SSH access
- Running long training jobs (hours to days)
- Want simple pricing with no egress fees
- Need persistent storage across sessions
- Require high-performance multi-node clusters (16-512 GPUs)
- Want pre-installed ML stack (Lambda Stack with PyTorch, CUDA, NCCL)
Key features:
- GPU variety: B200, H100, GH200, A100, A10, A6000, V100
- Lambda Stack: Pre-installed PyTorch, TensorFlow, CUDA, cuDNN, NCCL
- Persistent filesystems: Keep data across instance restarts
- 1-Click Clusters: 16-512 GPU Slurm clusters with InfiniBand
- Simple pricing: Pay-per-minute, no egress fees
- Global regions: 12+ regions worldwide
Use alternatives instead:
- Modal: For serverless, auto-scaling workloads
- SkyPilot: For multi-cloud orchestration and cost optimization
- RunPod: For cheaper spot instances and serverless endpoints
- Vast.ai: For GPU marketplace with lowest prices
Quick start
Account setup
- Create account at https://lambda.ai
- Add payment method
- Generate API key from dashboard
- Add SSH key (required before launching instances)
Launch via console
- Go to https://cloud.lambda.ai/instances
- Click "Launch instance"
- Select GPU type and region
- Choose SSH key
- Optionally attach filesystem
- Launch and wait 3-15 minutes
Connect via SSH
# Get instance IP from console
ssh ubuntu@<INSTANCE-IP>
# Or with specific key
ssh -i ~/.ssh/lambda_key ubuntu@<INSTANCE-IP>
GPU instances
Available GPUs
| GPU | VRAM | Price/GPU/hr | Best For |
|---|---|---|---|
| B200 SXM6 | 180 GB | $4.99 | Largest models, fastest training |
| H100 SXM | 80 GB | $2.99-3.29 | Large model training |
| H100 PCIe | 80 GB | $2.49 | Cost-effective H100 |
| GH200 | 96 GB | $1.49 | Single-GPU large models |
| A100 80GB | 80 GB | $1.79 | Production training |
| A100 40GB | 40 GB | $1.29 | Standard training |
| A10 | 24 GB | $0.75 | Inference, fine-tuning |
| A6000 | 48 GB | $0.80 | Good VRAM/price ratio |
| V100 | 16 GB | $0.55 | Budget training |
Instance configurations
8x GPU: Best for distributed training (DDP, FSDP)
4x GPU: Large models, multi-GPU training
2x GPU: Medium workloads
1x GPU: Fine-tuning, inference, development
Launch times
- Single-GPU: 3-5 minutes
- Multi-GPU: 10-15 minutes
Lambda Stack
All instances come with Lambda Stack pre-installed:
# Included software
- Ubuntu 22.04 LTS
- NVIDIA drivers (latest)
- CUDA 12.x
- cuDNN 8.x
- NCCL (for multi-GPU)
- PyTorch (latest)
- TensorFlow (latest)
- JAX
- JupyterLab
Verify installation
# Check GPU
nvidia-smi
# Check PyTorch
python -c "import torch; print(torch.cuda.is_available())"
# Check CUDA version
nvcc --version
Python API
Details, code examples and parameter tables: references/python-api.md. Read it when this step applies.
CLI with curl
List instance types
curl -u $LAMBDA_API_KEY: \
https://cloud.lambdalabs.com/api/v1/instance-types | jq
Launch instance
curl -u $LAMBDA_API_KEY: \
-X POST https://cloud.lambdalabs.com/api/v1/instance-operations/launch \
-H "Content-Type: application/json" \
-d '{
"region_name": "us-west-1",
"instance_type_name": "gpu_1x_h100_sxm5",
"ssh_key_names": ["my-key"]
}' | jq
Terminate instance
curl -u $LAMBDA_API_KEY: \
-X POST https://cloud.lambdalabs.com/api/v1/instance-operations/terminate \
-H "Content-Type: application/json" \
-d '{"instance_ids": ["<INSTANCE-ID>"]}' | jq
Persistent storage
Filesystems
Filesystems persist data across instance restarts:
# Mount location
/lambda/nfs/<FILESYSTEM_NAME>
# Example: save checkpoints
python train.py --checkpoint-dir /lambda/nfs/my-storage/checkpoints
Create filesystem
- Go to Storage in Lambda console
- Click "Create filesystem"
- Select region (must match instance region)
- Name and create
Attach to instance
Filesystems must be attached at instance launch time:
- Via console: Select filesystem when launching
- Via API: Include
file_system_namesin launch request
Best practices
# Store on filesystem (persists)
/lambda/nfs/storage/
├── datasets/
├── checkpoints/
├── models/
└── outputs/
# Local SSD (faster, ephemeral)
/home/ubuntu/
└── working/ # Temporary files
SSH configuration
Add SSH key
# Generate key locally
ssh-keygen -t ed25519 -f ~/.ssh/lambda_key
# Add public key to Lambda console
# Or via API
Multiple keys
# On instance, add more keys
echo 'ssh-rsa AAAA...' >> ~/.ssh/authorized_keys
Import from GitHub
# On instance
ssh-import-id gh:username
SSH tunneling
# Forward Jupyter
ssh -L 8888:localhost:8888 ubuntu@<IP>
# Forward TensorBoard
ssh -L 6006:localhost:6006 ubuntu@<IP>
# Multiple ports
ssh -L 8888:localhost:8888 -L 6006:localhost:6006 ubuntu@<IP>
JupyterLab
Launch from console
- Go to Instances page
- Click "Launch" in Cloud IDE column
- JupyterLab opens in browser
Manual access
# On instance
jupyter lab --ip=0.0.0.0 --port=8888
# From local machine with tunnel
ssh -L 8888:localhost:8888 ubuntu@<IP>
# Open http://localhost:8888
Training workflows
Single-GPU training
# SSH to instance
ssh ubuntu@<IP>
# Clone repo
git clone https://github.com/user/project
cd project
# Install dependencies
pip install -r requirements.txt
# Train
python train.py --epochs 100 --checkpoint-dir /lambda/nfs/storage/checkpoints
Multi-GPU training (single node)
# train_ddp.py
import torch
import torch.distributed as dist
from torch.nn.parallel import DistributedDataParallel as DDP
def main():
dist.init_process_group("nccl")
rank = dist.get_rank()
device = rank % torch.cuda.device_count()
model = MyModel().to(device)
model = DDP(model, device_ids=[device])
# Training loop...
if __name__ == "__main__":
main()
# Launch with torchrun (8 GPUs)
torchrun --nproc_per_node=8 train_ddp.py
Checkpoint to filesystem
import os
checkpoint_dir = "/lambda/nfs/my-storage/checkpoints"
os.makedirs(checkpoint_dir, exist_ok=True)
# Save checkpoint
torch.save({
'epoch': epoch,
'model_state_dict': model.state_dict(),
'optimizer_state_dict': optimizer.state_dict(),
'loss': loss,
}, f"{checkpoint_dir}/checkpoint_{epoch}.pt")
1-Click Clusters
Overview
High-performance Slurm clusters with:
- 16-512 NVIDIA H100 or B200 GPUs
- NVIDIA Quantum-2 400 Gb/s InfiniBand
- GPUDirect RDMA at 3200 Gb/s
- Pre-installed distributed ML stack
Included software
- Ubuntu 22.04 LTS + Lambda Stack
- NCCL, Open MPI
- PyTorch with DDP and FSDP
- TensorFlow
- OFED drivers
Storage
- 24 TB NVMe per compute node (ephemeral)
- Lambda filesystems for persistent data
Multi-node training
# On Slurm cluster
srun --nodes=4 --ntasks-per-node=8 --gpus-per-node=8 \
torchrun --nnodes=4 --nproc_per_node=8 \
--rdzv_backend=c10d --rdzv_endpoint=$MASTER_ADDR:29500 \
train.py
Networking
Bandwidth
- Inter-instance (same region): up to 200 Gbps
- Internet outbound: 20 Gbps max
Firewall
- Default: Only port 22 (SSH) open
- Configure additional ports in Lambda console
- ICMP traffic allowed by default
Private IPs
# Find private IP
ip addr show | grep 'inet '
Common workflows
Workflow 1: Fine-tuning LLM
# 1. Launch 8x H100 instance with filesystem
# 2. SSH and setup
ssh ubuntu@<IP>
pip install transformers accelerate peft
# 3. Download model to filesystem
python -c "
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained('meta-llama/Llama-2-7b-hf')
model.save_pretrained('/lambda/nfs/storage/models/llama-2-7b')
"
# 4. Fine-tune with checkpoints on filesystem
accelerate launch --num_processes 8 train.py \
--model_path /lambda/nfs/storage/models/llama-2-7b \
--output_dir /lambda/nfs/storage/outputs \
--checkpoint_dir /lambda/nfs/storage/checkpoints
Workflow 2: Batch inference
# 1. Launch A10 instance (cost-effective for inference)
# 2. Run inference
python inference.py \
--model /lambda/nfs/storage/models/fine-tuned \
--input /lambda/nfs/storage/data/inputs.jsonl \
--output /lambda/nfs/storage/data/outputs.jsonl
Cost optimization
Choose right GPU
| Task | Recommended GPU |
|---|---|
| LLM fine-tuning (7B) | A100 40GB |
| LLM fine-tuning (70B) | 8x H100 |
| Inference | A10, A6000 |
| Development | V100, A10 |
| Maximum performance | B200 |
Reduce costs
- Use filesystems: Avoid re-downloading data
- Checkpoint frequently: Resume interrupted training
- Right-size: Don't over-provision GPUs
- Terminate idle: No auto-stop, manually terminate
Monitor usage
- Dashboard shows real-time GPU utilization
- API for programmatic monitoring
Common issues
| Issue | Solution |
|---|---|
| Instance won't launch | Check region availability, try different GPU |
| SSH connection refused | Wait for instance to initialize (3-15 min) |
| Data lost after terminate | Use persistent filesystems |
| Slow data transfer | Use filesystem in same region |
| GPU not detected | Reboot instance, check drivers |
References
- Advanced Usage - Multi-node training, API automation
- Troubleshooting - Common issues and solutions
Resources
- Documentation: https://docs.lambda.ai
- Console: https://cloud.lambda.ai
- Pricing: https://lambda.ai/instances
- Support: https://support.lambdalabs.com
- Blog: https://lambda.ai/blog
Related skills
modal-serverless-gpu: Serverless GPU cloud platform for running ML workloads.skypilot-multi-cloud-orchestration: Multi-cloud orchestration for ML workloads with automatic cost optimization.modal: Modal is a serverless cloud platform for running Python on demand, including on-demand GPUs.
Trigger Prompts & Evaluation Corpus
These evaluation prompts test that the agent correctly selects lambda-labs-gpu-cloud for relevant tasks and leaves it out for near misses.
Should trigger (3)
- should trigger
“How do I attach a persistent filesystem to a Lambda instance? I launched one yesterday and can't see it.”
expects Explains that filesystems must be attached at launch time (console selection or `file_system_names` in the API launch request), must be in the same region as the instance, and that an already-launched instance cannot have one attached, so the user must relaunch. Mentions the filesystem mount location and keeping data/checkpoints there.
- should trigger
“I'm fine-tuning Llama-3 70B with FSDP using train.py and need a box with 8x H100 and SSH access for about 3 days. My checkpoints are ~140GB each and I don't want to lose them if I terminate. Walk me through launching it with curl against the Lambda API and getting my dataset onto it.”
expects Gives steps to list instance types, launch an 8x H100 instance via the Lambda Cloud API with an SSH key name, region, and a same-region filesystem in `file_system_names`, then SSH in and write checkpoints/data to the filesystem. Notes that there is no auto-stop so the instance must be terminated manually, and that multi-GPU launch takes ~10-15 minutes.
- should trigger
“I need to train across 128 GPUs with fast interconnect for a pretraining run. What are my options for a managed Slurm cluster with InfiniBand?”
expects Recommends Lambda 1-Click Clusters (16-512 H100/B200 GPUs, Slurm, Quantum-2 400 Gb/s InfiniBand, GPUDirect RDMA, pre-installed NCCL/PyTorch). Notes that node-local NVMe is ephemeral and Lambda filesystems should be used for persistent data, and shows a multi-node launch via Slurm.
Near misses (2)
- near miss
“I want to deploy my fine-tuned model as an autoscaling HTTPS endpoint that scales to zero when idle and only bills per request. How should I set that up?”
expects Recognizes this is a serverless, auto-scaling deployment need and points to Modal (modal-serverless-gpu) rather than Lambda Labs dedicated instances, which have no auto-stop or scale-to-zero.
- near miss
“I have credits on AWS, GCP and Azure. Write a config that runs my training job on whichever cloud has the cheapest spot H100 and automatically recovers if it's preempted.”
expects Identifies a multi-cloud spot orchestration task and uses SkyPilot (task YAML with resources, spot recovery) instead of Lambda Labs-specific instance setup.