Your developers are building AI features. Your content team is generating assets. Who can use which models, and how do you keep spending under control? Compare ModelGo and Portkey on model access, team permissions, and budgets before choosing your AI gateway.
ModelGo's enterprise AI gateway brings model access and management into one place, with routing and failover, team permissions, budget controls, and request tracing. Use models supplied through ModelGo, with no additional platform service fee.
Enterprise AI gateway features compared
| Feature | Portkey | ModelGo |
|---|---|---|
| Model access | Connect existing provider credentials through the Model Catalog. | Access models supplied through ModelGo using your ModelGo account and API key. |
| Platform service fee | Free developer plan; paid hosted production and enterprise plans. Open-source self-hosting is also available. | No platform service fee. Model usage is billed at the rates that apply to the selected model. |
| Routing and failover | Fallbacks, retries, load balancing, and other routing strategies. | Gateway routing and failover help reduce disruption when a route fails. |
| Existing applications and tools | SDKs, compatible APIs, and coding tool integrations. | Compatible APIs and native protocols for supported models, with setup guides for applications, coding tools, and video workflows. |
| Asynchronous generation jobs | Video job submission and status checks through supported providers. | Create video jobs, check their status, and retrieve results for product content and video production workflows. |
| Projects and workspaces | Workspaces with separate team resources and usage policies. | Create workspaces by project or business function. Manage members, API keys, model access, and request records within each workspace. |
| Team access | Roles and scoped API keys, with centrally managed access revocation. | Assign roles at the enterprise, workspace, and member levels. Revoke unused keys when people or projects change. |
| Model restrictions | Configure access to providers and models. | Set model access policies for workspaces or API keys so each application uses approved models. |
| Member and application budgets | Budget limits for developers, teams, workspaces, and other supported scopes. | Set budgets for members and API keys. Requests that exceed configured limits are blocked. |
| Cost tracking and reconciliation | Request costs and usage, with attribution by team, project, or user. | Review usage and costs by workspace, model, member, or API key to identify which project, person, or application incurred the spend. |
| Request troubleshooting | Logs, request traces, cost data, latency, and other observability features. | Review request IDs, status, latency, and error details to investigate application calls and generation jobs. |
| Request and response logging | Request and response logs and enterprise data controls; options vary by plan and deployment. | Enable input and output content logging per workspace. Configure log forwarding separately. |
| Administrative audit logs | Audit records for organization access, configuration, and policy actions. | Record key administrative changes to help investigate configuration, access, budget, and policy updates. |
Enterprise teams can apply for up to $200 in trial credits to test their chosen models, access settings, and budgets.
Teams and use cases
| AI application, SaaS, and development teams | E-commerce content teams | Short-form drama and animation teams |
|---|---|---|
| Model access and troubleshooting: Connect existing applications, use fallback routing, and inspect request logs when calls fail. | Generation status and asset retrieval: Check job status and retrieve completed assets for your content workflow. | Project and team access: Organize production in workspaces, assign model access to project members, and manage usage by project. |
| Application budgets and costs: Set spending limits for applications and review costs by project, model, and API key. | Content production costs: Review usage and costs by project, model, and team member. Set budgets for production staff or generation tools. | Production budgets and records: Set member and API key budgets, review project spend, and retain request records for troubleshooting. |
Other teams can contact ModelGo to discuss their requirements. More use cases will be added over time.
FAQ
How does ModelGo differ from Portkey?
Portkey's standard setup connects your existing model provider credentials. ModelGo supplies models directly, so teams can access supported models through a ModelGo account without setting up a separate account for each provider. ModelGo charges no additional platform service fee.
Do I need separate accounts with model providers?
No. Access models supplied through ModelGo with your ModelGo account and API key. Model availability and API requirements vary by model.
Does ModelGo charge a platform service fee?
No. ModelGo does not charge an additional platform service fee. Model usage is still billed at the rates that apply to your selected model.
Can I cap spending for a team member or application?
Yes. Set budgets for members and API keys, with requests blocked according to the configured limits. Assign separate keys to people or applications, then review their model access, usage, and costs. See the enterprise management overview.
Does one API key mean every model uses the same API?
No. Your ModelGo API key authenticates requests, but models can use different endpoint paths, parameters, and response formats. Follow the model catalog and the relevant protocol documentation. See the integration overview.
Can I keep my existing SDKs, coding tools, and video workflows?
For supported compatible APIs, setup usually involves changing the API address, API key, and model name. Coding tools and video workflows should follow the documentation for the specific tool, model, and protocol. Check the tool setup guides and video API documentation for supported features.
Can we control whether prompts and responses are logged?
Yes. Workspace settings control input and output content logging and display. Log forwarding is configured separately. These settings apply to ModelGo's logs; upstream providers have their own data retention policies. See the logging and data controls.
How do we start an enterprise trial?
Contact ModelGo with your use case, target models, and expected usage. Apply for an enterprise trial and up to $200 in trial credits to evaluate model access, permissions, budgets, and request records. The ModelGo team will confirm the credit amount and trial arrangements.
Start building with ModelGo
Explore the docs and connect your first model.