Context Engine
- Online queries
- Offline queries
- Resolvers
- Chalk SQL
Billed by machine type
Each Context Engine machine type has one hourly rate.
Machines × machine type rate × runtime = cost
Chalk pricing
One Platform, Usage-Based Pricing
Pay for the compute you use, down to the second.
Standard rates
Compute
| CPU and memory | Preemptible |
|---|---|
| CPU | $0.0249 |
| Memory | $0.0080 |
Preemptible rates apply unless you request non-preemptible workloads, which run for up to 24 hours.
| GPUs | |
|---|---|
| NVIDIA H100 (80 GB) | $3.92 |
| NVIDIA H200 (141 GB) | $4.55 |
| NVIDIA B200 (180 GB) | $6.23 |
| NVIDIA B300 (288 GB) | $7.10 |
| NVIDIA L4 (24 GB) | $0.79 |
| NVIDIA L40S (48 GB) | $1.92 |
| NVIDIA A100 (40 GB) | $2.10 |
| NVIDIA A100 (80 GB) | $2.49 |
| NVIDIA A10 (24 GB) | $1.09 |
| Storage and network | |
| Volumes | $0.09 |
| Network egress | $0.04 |
| Network ingress | No charge |
Context Engine machine types
| Machine type | Rate |
|---|---|
| micro2 vCPU · 4 GiB | $0.38 / hr |
| small4 vCPU · 16 GiB | $0.95 / hr |
| medium8 vCPU · 32 GiB | $1.90 / hr |
| large16 vCPU · 64 GiB | $3.80 / hr |
| xlarge32 vCPU · 128 GiB | $7.60 / hr |
AI usage
| AI provider |
|---|
| ChalkInput $0.10 · Cached input $0.01 · Cache write $0.125 · Output $0.50 / 1M tokens |
| Your own providerBilled by your provider |
Chalk Assistant uses Chalk as its default AI provider, or you can connect your own. Reasoning tokens bill as output. Chalk adds no fee to tokens from your own provider in Chalk Assistant, Model Gateway, or MCP Gateway.
Billing
How each product is billed
Runtime
- Model serving
- Sandboxes
- Functions
- Scaling groups
Billed by requested resources
Pay for the vCPU, memory, and GPUs each container requests, by the second, at the rates above.
(vCPU × CPU rate + GiB × memory rate + GPUs × GPU rate) × runtime = cost
Learning
- Model training
- Agent evals
- Notebooks
Billed by where it runs
Model training, notebooks, and other container workloads bill at Runtime rates.
Evals that run on offline query infrastructure bill at Context Engine rates.
Start building
Pay as you go
Get started today and only pay for what you use, at per-second granularity. No up-front costs.
Learn how pricing worksSave more with
Committed use contracts
Larger commitments can unlock greater discounts and self-hosted deployment options.
Contact usPlans
Compare pricing tiers
| Feature | Standard | Enterprise | Business Critical |
|---|---|---|---|
| Pricing | List rates ↑ | Contact us | Contact us |
| Support and communication | |||
| Helpdesk support (email and ticket) | Included | Included | Included |
| Access to Chalk Community Slack | Included | Included | Included |
| Standard SLA | Included | Included | Included |
| Dedicated Slack channel with Chalk Team | Not included | Included | Included |
| Phone and video support | Not included | Included | Included |
| Priority response SLA | Not included | Not included | Included |
| Compute | |||
| Custom non-preemptible durations | Not included | Included | Included |
| Security and compliance | |||
| IP allowlisting | Not included | Included | Included |
| Custom RBAC | Not included | Included | Included |
| Audit trails | Not included | Included | Included |
| Advanced SSO (Okta, AD) | Not included | Included | Included |
| Observability and integrations | |||
| Metrics export (OpenMetrics) | Not included | Included | Included |
| Customer success | |||
| Priority feature requests | Not included | Not included | Included |
| Onboarding training sessions | Not included | Not included | Included |
| Code review for integrations and deployment of features + resolvers | Not included | Not included | Included |
| Embedded forward-deployed engineering | Not included | Not included | Add-on |
Talk to an engineer
Find the right tier for your workload.
Tell us what you’re building and we’ll help you estimate usage and choose a plan.