PRICING

Plans to power

your AI transformation

We work with teams of all sizes to scale AI usage

Enterprise

Tailored

Every AI transformation is different

Let's chat
Connect

  • SaaS catalog integrations

  • Custom server creation

  • Code mode efficiency

  • Serverless hosting

Secure

  • OAuth 2.1 proxy + DCR

  • Automatic provisioning

  • Scoped RBAC controls

  • Full audit trail

Observe

  • Real-time logs

  • Distributed tracing

  • Session tracking

  • User analytics

Support

  • SLA-backed support

  • Dedicated FDE support

  • Concierge onboarding

  • Dedicated support channel

Select a plan:

Tool creation

OpenAPI import
Default toolset creation
Custom toolset creation

Tool curation

Custom tool creation
Response filtering
Multi-tool workflow builder
Custom prompt creation
LLM playground testing

Server authentication

Provided API credentials
Standard OAuth
BYO authorization server
OAuth 2.1 proxy + DCR

Server management

Public / private server control
Custom domain support
Version tracking

Security

SSO
Audit logs

Support

Support channels
Priority Slack + Teams + chat + email
Early feature access
Service SLAs
Support SLAs
Concierge onboarding
Legal + security reviews

Frequently AskedQuestions

What defines a request?
A request is a tool call made by an LLM to your MCP server.
What's a tool?
A tool is a function surfaced in the MCP server for use by the LLM. Think of it as the MCP version of an API operation.
What's a custom tool?
It is possible to have a tool call that orchestrate multiple sub tool calls. This is what we call a custom tool.
How do custom tools impact the number of requests?
It is possible to have a tool call that will orchestrate multiple sub tool calls. Each tool call counts as a request. For example if get_recent_account_activity uses sub tools: get_account and get_activity. That would be counted as 3 tools calls.
Why do I need multiple MCP servers?
You don't necessarily need multiple MCP servers, but it's incredibly useful! LLMs tend to struggle as the number of tools increases. Therefore, you're better off having more servers with fewer tools. Think of it like having different toolboxes for different jobs.
How many tools can an MCP server include?
While there's no hard technical limit, less is more when it comes to MCP servers. LLM performance starts to drop significantly when an MCP server has 30-40 tools. LLMs work best with focused, curated toolsets rather than access to every possible tool.
Do LLM playground requests count toward usage?
Yes they do.
Supported:
ModelContextProtocol
Select a billing commitment

Free

$0

Features

  • SDK & MCP generation

  • Auto-publish to package manager

  • Access to our OpenAPI toolchain

What's included

  • 250 operations / language

  • Community Slack support

  • License for 1 language

Start building

Business

$600

/ month per language

Features

  • OAuth support

  • Webhooks and SSE

  • Docsite integration

What's included

  • 250 operations / language

Start free trial

Enterprise

Tailored

Business features +

  • Dedicated Slack support

  • Concierge onboarding + migration

  • SLAs for service and support

  • SSO and audit logs

  • Customizable to your requirements

Let's chat

Free

$0

Features

  • SDK & MCP generation

  • Auto-publish to package manager

  • Access to our OpenAPI toolchain

What's included

  • 250 operations / language

  • Community Slack support

  • License for 1 language

Start building

Business

$720

/ month per language

Features

  • OAuth support

  • Webhooks and SSE

  • Docsite integration

What's included

  • 250 operations / language

Start free trial

Enterprise

Tailored

Business features +

  • Dedicated Slack support

  • Concierge onboarding + migration

  • SLAs for service and support

  • SSO and audit logs

  • Customizable to your requirements

Let's chat
Select a plan:

OpenAPI tooling

Spec validation
Custom linting rules
OpenAPI overlay editing
OpenAPI web editor with linting
OpenAPI change history
API linting reports
API audit log

SDK features

OpenAPI 3.0 & 3.1 support
Type-safety
Data validation
File streaming downloads
File streaming uploads
Retries
Semantic versioning
Publish to package managers
SDK changelog
Generate React queries
Pagination
SDK testing
JSONL support
Server-sent events
Webhook Support
Custom SDK migration

Code samples

Code sample generation
Code samples API

Custom code

Bring your own dependencies
Runtime hooks
Custom code regions

Authentication

Basic, Bearer, API key
OAuth2.0
Custom security schemes

Security

SSO
Audit logs

Support

Support channels
Community
Service SLAs
Early feature access
Support SLAs
Concierge onboarding

Frequently AskedQuestions

How do operations and languages affect pricing?
Pricing is based on the languages you license and the number of unique API operations you generate SDKs for. When you license a language, you can generate any number of SDKs in that language. However, all SDKs for the same language share a single operation limit. Operations are counted by unique operationId values in your OpenAPI spec. On the Business plan, this limit is 250 unique operations per licensed language. For example, if you license TypeScript on the Business plan, you can generate multiple TypeScript SDKs (such as internal and external SDKs), as long as the total number of unique operations across all TypeScript SDKs does not exceed 250.
What counts as an operation?
An operation is a GET, POST, PATCH, DELETE operation in your API interface.
Am I charged for endpoints I exclude from my SDK?
No. You're only charged for the methods you choose to include. We make it easy to exclude any endpoints you don't want in your SDK.
What if my API grows and I exceed my usage limits?
Congrats on the growth! If you exceed your usage limits, our team will check in at the end of the quarter to discuss upgrading your plan.
What happens if we stop working together?
You own 100% of the code you generate. You'll keep everything you've created. You just won't be able to generate new code through the Speakeasy platform.

AI everywhere.

Control here.

Talk to usLearn more