Skip to main content
Move your existing LLM integration to NexLLM while keeping your application’s core request flow. NexLLM provides an OpenAI-compatible API for accessing GPT, Claude, and Gemini models. For a basic Chat Completions integration, migration starts with updating the base URL and API key, then verifying model identifiers and any gateway-specific settings.

​
Pick your starting point

Match what you’re using today to the guide that walks through the swap.

OpenRouter

Update your base URL, configure your NexLLM key, and verify model IDs.

Helicone

Review your Helicone proxy configuration and move your calls to NexLLM.

Portkey

Review virtual keys, headers, and routing before switching to NexLLM.

Merge Gateway

Update your gateway configuration and validate requests with NexLLM.

Vercel AI Gateway

Configure your AI SDK or OpenAI client to send requests to NexLLM.

TensorZero

Review your gateway features before moving inference calls to NexLLM.

LiteLLM

Review model aliases and proxy settings before switching to NexLLM.

​
Another OpenAI-compatible provider or gateway

Use the common migration checklist below. Keep the existing client where practical, but review custom headers, authentication, model aliases, and gateway features individually. If your application depends on provider-specific routing, fallback policies, caching, or privacy controls, treat those requirements as separate migration tasks—not just configuration changes.

​
What every migration has in common

Use the following configuration for NexLLM’s OpenAI-compatible endpoints. The Chat Completions path below is relative to the SDK base URL. (docs.nexllm.ai) Do not append /v1 twice. The complete Chat Completions endpoint is https://www.nexllm.ai/v1/chat/completions. (docs.nexllm.ai)

​
1. Configure your credentials

Store your NexLLM key using your existing environment configuration or secrets manager. Use placeholders in committed example files, and never paste real credentials into documentation or migration conversations.

​
2. Confirm model access

NexLLM associates each API key with a channel group. That group controls model access and the pricing ratio applied to usage. Choose the appropriate group before validating your integration. (docs.nexllm.ai)

​
3. Discover exact model identifiers

Query the Models endpoint using the same key your application will use: bash
Use the returned data[].id values in inference requests. The catalog is key-specific, so different keys may return different models. (docs.nexllm.ai) Do not silently substitute a different model or upgrade a model version during migration. Confirm any identifier mapping before changing production calls.

​
4. Update and test your client

Change the client’s base URL and credentials, apply approved model mappings, and test a minimal request. Review gateway-specific headers and request fields before removing them. Test streaming, tools, structured outputs, and multimodal inputs separately when your application requires them. Keep rollback configuration available securely until validation is complete.

​
What you get with NexLLM

​
Multiple model families through one API

Access GPT, Claude, and Gemini models through a unified API base URL, subject to your key’s model access. (docs.nexllm.ai)

​
An OpenAI-compatible integration

Retain the OpenAI SDK for supported requests rather than introducing a NexLLM-specific client library. (docs.nexllm.ai)

​
Channel-group access and pricing controls

Select the channel group appropriate for your integration’s model access and pricing requirements. Each key belongs to one group. (docs.nexllm.ai)

​
Streaming responses

Chat Completions supports Server-Sent Events with stream: true, allowing applications to display generated text incrementally. (docs.nexllm.ai)

​
Additional API formats

Beyond Chat Completions, NexLLM documents embeddings, image and audio endpoints, an OpenAI-compatible Responses endpoint, and an Anthropic-format Messages endpoint. Verify the chosen model’s capabilities before adopting a different API format. (docs.nexllm.ai)

​
Privacy and compliance considerations

Treat privacy requirements as an explicit migration checkpoint. Before moving sensitive workloads, confirm:
  • Request and response logging practices.
  • Upstream provider data-retention policies.
  • Any required regional or provider restrictions.
  • Whether fallback behavior preserves those restrictions.
  • Applicable contractual requirements.
Do not assume that a privacy setting from your previous gateway transfers to NexLLM. Obtain confirmation of the required guarantees before moving affected traffic.

​
Migration checklist

  • Configure a NexLLM API key securely.
  • Confirm its channel group.
  • Discover models using the target key.
  • Approve model mappings.
  • Update client configuration.
  • Review gateway-specific behavior.
  • Test required application features.
  • Validate deployment secrets and rollback.
  • Retire unused credentials after a successful rollout.

​
Feedback

When reporting a migration issue, include the previous integration, SDK version, model identifier, endpoint, and a sanitized error or minimal reproduction. Never include API keys, authorization headers, or sensitive request content.