Skip to steps

NEXO

Gateway guide

OpenAI-compatible · key per request · streaming supported

Call the gateway in three steps

Same shape as the OpenAI chat API. Only the base URL and the key change.

Steps

  1. 1. Get an API key

    Ask the gateway operator for a key. Send it on every request as Authorization: Bearer $NEXO_KEY. Without it you get 401 missing_api_key; an expired one returns 402 key_expired.

  2. 2. Point at the base URL

    Pick a live model id from the status page. Anything marked Down there will fail the same way here.

  3. 3. POST a chat completion

    Choose a model, pick a tab, copy the snippet. Add "stream": true to the body for server-sent events.

Try it

Loading model list…


        

      


      

        

          
Use it from your agent

          
Same key, same base URL — per tool

        


        

          
        


        

          

            

            
          

          

        

        

          Replace sk-your-client-key with your real NEXO client key.
          Only the Hermes tab follows this repo's verified skill docs; the rest follow each tool's
          standard OpenAI-compatible pattern — tell us if one drifts.
        

      


      

        
Errors you may meet

        

          

            
401 · 402

            

              Missing, invalid, or expired key. Check the header and ask for a fresh key.
            

          

          

            
429

            

              Rate limit: 30 requests per minute per key. Back off and retry.
            

          

          

            
502 · 503

            

              Upstream trouble. Retry once; if it persists, check the status page.
            

          

          

            
Streaming stalls

            

              Read the SSE stream to data: [DONE].
              The first chunk may carry no text yet.