🔑 LLM Access Portal

Sign in

Accounts are provisioned by an owner — ask an owner to create yours.

Set a new password

Workspaces

API base URLhttps://llm.dogfooddevsecure.com/v1
Workspace budget caps the team in total; per-key caps each key on its own, so one runaway client can't drain the workspace.
Models for this workspace — none checked = all free models. Paid cloud models must be ticked on purpose, and only an owner/admin can grant them:
—
all models selected

API keys

—

Available models

—

Configure OpenClaw

Pick a model, then copy the commands to point your OpenClaw at the LLM Portal. Run them in Terminal (Mac/Linux) or in WSL (Windows). Paste your own API key (from “API keys”) — it stays in your browser.
(select a model)
Use only models your key/workspace allows.

Pending requests

—

Users (owner)

—

Accounts

—

Models available to non-owners (none checked = all)

—
all models selected

Model access sync

Pushes each workspace's intended models to its LiteLLM team. Run this once after enabling paid models — workspaces created before then are still unrestricted at the gateway, meaning their existing keys can reach the paid models. Also run it after adding a new free model, since explicit lists don't pick one up on their own. Always previews first.

      

      
Monitoring

        

          
          
          
          
          
          
          
        

        

          
—

          

          

          

          

          

          
Access

          
—

        

      

      
Activity

        

          
          
          
          
          
        

        

          

          

          

            
By workspace
—

            
By model
—

          

          

            
            
            
          

          

            
—

          

          

            
—

          

          

            

              
              
              
              
              
              
            

            
—

            

              
              
              
            

          

        

      

      
Storage

        
Storage backends for the Creative Studio system, via the Studio storage-admin API (the Portal backend holds the admin key).

        

          
          
          
          
        

        
—

      

      
Agents

        
Headless ops agents (Buzz members) running on the model hosts. They phone home every cycle; model changes apply on their next sync. gateway: models route via LiteLLM (needs a portal key on the agent's box; paid models bill AWS) — local: models use the box's own Ollama.

        

          
          
        

        
—

        

        
Register agent
          

            
            
            
            
            
            
          

        

      

      
Features & Tasks

        

          
          
          
          
          
        

        
—

      

    

   

  


  

    
Copy this now — it won't be shown again