Research machines
Work from anywhere on a private research instance pre-configured with a multi-kernel Jupyter server, VS Code, and SSH with root access
The trading workflow usually starts with researching an idea. The streamline the development workflow, you can spin up a VM, with dedicated and guaranteed resources, preloaded with Jupyter server and VS Code - allowing you to conduct your work from anywhere.
Since research instances are work machines, it's generally recommended that you create your machine in a data center that's closest to you.
We currently have instances in 6 data centers around the world, with more that can be made available upon request:
us-east-1- Newark, United Statesus-east-2- New York City, United Statesus-west-1- San Francisco, United Stateseu-central-1- Frankfurt, Germanyeu-west-1- London, United Kingdomap-southeast-1- Singapore
Servers in India, Canada, and the Netherlands can be added upon request.
To create an instance using the web console, simply click the Create Instance button on the instances page, and enter your machine's name, configuration, and location
Within a minute or two, your machine will be ready to use:
Once its ready, use the auto-generated password to SSH into the server, launch Jupyter lab/notebook, or VS Code.
Remember - this machine is dedicated to you, so you can install anything to make yourself feel at home.
Research instances are priced by the hour and billed by the minute. The idea is that not to spin up a machine and keep it running 24/7, but rather to use them on-demand and kill them when you don't need them.
However, when you kill a machine you lose all your work.
We address this issue by providing the following solutions:
Block storage volumes are SSD-based volumes you can attach to your research instances for data persistency between sessions.
You get 10GB/mo of free storage with your account (1GB/mo for the pay-as-you-go plan) and you can split it between different data centers if you so want (5GB in the EU and 5GB in the US). Additional storage is priced at $0.14 - $0.25 per GB per month, depending on usage.
You should save any work product (files, ML models, etc) you'd like to use in the future to your volume, and, when you create a new instance, simply attach this volume to the new instance for instance access to your files.
NOTE: Block storage volumes can only be attached to research instances within the same data center!
Coming Soon
To make your life even easier, we'll soon be launching Research Snapshots. This feature will create an image of your research instance machine, and use that image to create a new instance later.
Snapshots have the added benefit of saving your entire configuration - so any modifications you made, including installed software, settings, themes, etc - will be moving with you to the new instance.
Snapshots will be charged based on the actual disk usage! For example - if your instance has a 1TB disk, but you've only actually used 10GB - your snapshot will be 10GB, and will cost ~$1.5/mo to store.
Size | Price/mo * | Price/hr |
|---|---|---|
First 10GB | ~$0.15 / GB | $0.00020 / GB |
10GB - 1TB | ~$0.10 / GB | $0.00014 / GB |
1TB+ | ~$0.05 / GB | $0.00007 / GB |
Virtual machines that offers a good balance of memory and vCPUs.
Type | vCPUs | Memory | SSD | Price/hr |
|---|---|---|---|---|
b.small | 4 | 16GB | 50GB | $0.45 |
b.medium | 8 | 32GB | 100GB | $0.85 |
b.large | 16 | 64GB | 200GB | $1.75 |
b.xlarge | 32 | 128GB | 400GB | $3.50 |
optimized VMs with dedicated CPU for workloads that rely on CPU more than RAM.
Type | vCPUs | Memory | SSD | Price/hr |
|---|---|---|---|---|
c.small | 4 | 8GB | 100GB | $0.45 |
c.medium | 8 | 16GB | 200GB | $0.85 |
c.large | 16 | 32GB | 400GB | $1.75 |
c.xlarge | 32 | 64GB | 800GB | $2.75 |
Virtual machines with 8GB of memory for each vCPU for RAM-intensive research.
Type | vCPUs | Memory | SSD | Price/hr |
|---|---|---|---|---|
m.small | 4 | 32GB | 100GB | $0.55 |
m.medium | 8 | 64GB | 200GB | $1.65 |
m.large | 16 | 128GB | 400GB | $2.55 |
m.xlarge | 32 | 256GB | 800GB | $4.75 |
Coming Soon
GPU-optimized VMs for AI/ML-focused research and machine learning model training. Accelerated by the NVIDIA Quadro RTX 6000, harnessing the power of CUDA, Tensor, and RT cores.
Type | vCPUs | Memory | SSD | GPUs | CUDA* | Tensor* | Price/hr |
|---|---|---|---|---|---|---|---|
g.small | 8 | 32GB | 60GB | 1 (24GB) | 4,066 | 576 | $0.45 |
g.medium | 16 | 64GB | 1.2TB | 2 (48GB) | 9,216 | 1,152 | $0.85 |
g.large | 20 | 96GB | 1.9TB | 3 (72GB) | 13,824 | 1,728 | $1.75 |
g.xlarge | 24 | 128GB | 2.5TB | 4 (96GB) | 18,432 | 2,304 | $3.50 |
* CUDA and Tensor refer to the number of cores
Last modified 1yr ago