Pricing · Compare plans
Compare DeepintShield features by plan
This page compares DeepintShield features across the Developer, Cloud Team, Cloud Business and Enterprise plans, using the plan limits published in our docs. Every plan includes the AI gateway, built-in guardrails and the MCP registry. Higher plans add ML detectors, Agentic Security, Hallucination Defense and deployment in your own cloud.
How to read the DeepintShield features tables
Each plan includes the features of the plan to its left; hosting and limits differ. The Enterprise column covers Enterprise VPC and Enterprise Self-Hosted. Prices are on the pricing page, and plan gating follows the billing docs.
- Yes: included in the plan price
- No: not available on that plan
- Ask us: available on request; we confirm the details for your plan on a call
- Add-on: a paid service, priced in the services table
- A number: the limit on that plan
Gateway, guardrails, agents and MCP
Swipe sideways to see every column →
| Feature | Developer | Cloud Team | Cloud Business | Enterprise |
|---|---|---|---|---|
| AI gateway | ||||
| One OpenAI-compatible API for 29 built-in model providers | Yes | Yes | Yes | Yes |
| Governed requests a month | 50,000 | 500,000 | 3,000,000 | 10M or 25M |
| Adaptive load balancing with circuit breaker | No | Yes | Yes | Yes |
| Prompt Playground | Unsaved only | Saved and versioned | Saved and versioned | Saved and versioned |
| Runtime guardrails | ||||
| Built-in guardrails, including OWASP LLM Top-10 cards | Yes | Yes | Yes | Yes |
| Guardrail policies | 1 | Unlimited | Unlimited | Unlimited |
| ML detector sidecar | No | Yes | Yes | Yes |
| Safety providers: AWS Bedrock Guardrails, Azure AI Content Safety, Google Model Armor, your webhook | No | Yes | Yes | Yes |
| RAG Security live evaluation | No | Yes | Yes | Yes |
| Hallucination Defense: five detectors plus an overall risk score | No | No | Yes | Yes |
| Agents and MCP | ||||
| MCP registry, Agent Mode and Code Mode | Yes | Yes | Yes | Yes |
| MCP servers | 1 | 25 | 50 | Unlimited |
| MCP tool-result caching (opt-in) | No | Yes | Yes | Yes |
| Agentic Security: Allow, Deny or Require approval on tool calls | No | No | Yes | Yes |
| Agent identity and enrolment | No | No | Yes | Yes |
| Pinned tool definitions with re-approval on change | Ask us | Ask us | Yes | Yes |
| AIBOM export, signed when a signing key is configured | No | No | Yes | Yes |
Governance, identity, observability and audit
Swipe sideways to see every column →
| Feature | Developer | Cloud Team | Cloud Business | Enterprise |
|---|---|---|---|---|
| Governance and identity | ||||
| Virtual keys with budgets and rate limits | Yes | Yes | Yes | Yes |
| Teams and customers with their own budgets | No | Yes | Yes | Yes |
| Per-model budgets and rate limits | No | Yes | Yes | Yes |
| Relationship-based access with OpenFGA | No | No | Yes | Yes |
| TOTP multi-factor authentication | Yes | Yes | Yes | Yes |
| SSO over OIDC: Microsoft Entra ID, Okta, Google Workspace, Auth0 | Ask us | Ask us | Ask us | Yes |
| Role-based access with custom roles | Ask us | Ask us | Ask us | Yes |
| SCIM sync with Microsoft Entra ID | Ask us | Ask us | Ask us | Ask us |
| Console seats | 2 | 10 | 30 | 50 |
| Protected identities | Ask us | Ask us | Ask us | 500 or 1,000 |
| Observability & Audit | ||||
| Request logs with tokens, cost and latency, no logging surcharge | Yes | Yes | Yes | Yes |
| Agent decision ledger: append-only, hash-chained | No | No | Yes | Yes |
| Audit logs with a SHA-256 hash chain and Verify chain | Ask us | Ask us | Ask us | Yes |
| Log exports to S3, GCS, Azure Blob, Snowflake, Redshift, BigQuery | Ask us | Ask us | Ask us | Yes |
| OpenTelemetry and Prometheus metrics | Ask us | Ask us | Ask us | Ask us |
| Durable webhooks | Ask us | Ask us | Ask us | Ask us |
Cost, deployment and support
Swipe sideways to see every column →
| Feature | Developer | Cloud Team | Cloud Business | Enterprise |
|---|---|---|---|---|
| Cost Optimization | ||||
| Bring your own keys, 0% markup on model usage | Yes | Yes | Yes | Yes |
| Five baseline techniques, including provider prompt caching and request coalescing | Yes | Yes | Yes | Yes |
| Semantic caching | No | Yes | Yes | Yes |
| Cascade routing, prompt compression, reasoning-effort controls and three more techniques | No | Yes | Yes | Yes |
| Agentic Cache reporting | No | No | Yes | Yes |
| Deployment and support | ||||
| Fully managed DeepintShield Cloud | Yes | Yes | Yes | Ask us |
| Hybrid: data plane in your GKE, AKS or EKS cluster | No | No | No | Enterprise VPC |
| Self-hosted, including air-gapped | No | No | No | Enterprise Self-Hosted |
| Production clusters | Managed | Managed | Managed | One (VPC) or up to three (Self-Hosted) |
| Implementation service | No | No | No | Add-on |
| 24x7 premium support with a named technical account manager | Ask us | Ask us | Ask us | Add-on |
| Air-gapped delivery | No | No | No | Add-on (Self-Hosted) |
Services and add-ons
Services are priced separately from the plan, in USD and excluding taxes.
| Service | Applies to | Price |
|---|---|---|
| Implementation (one-time) | Enterprise VPC / Self-Hosted | Ask us |
| 24x7 premium support with a named technical account manager | Per year | Ask us |
| Air-gapped delivery | Self-Hosted only, per year | Ask us |
Not sure which plan fits?
Start free on the Developer plan and upgrade when you need more, or talk to us about Enterprise in your own cloud.
