faire un abonnement vast.ai
installer un serveur ollama
faire tourner plusieirs modeles successivement.
faire tourner plusieurs modeles en meme temps.


ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIDw6NFpHs1rqxwoi4Dmy9eQX4wJdQ1EOprD49qO8ltql nicolas b@DESKTOP-KPPH6RI

ssh -p 13777 root@ssh2.vast.ai -L 8080:localhost:8080


ssh -i "C:\Users\Nicolas B\.ssh\id_vast_demo" -p 17725 root@ssh2.vast.ai -L 8080:localhost:8080

ssh -i "C:\Users\Nicolas B\.ssh\id_vast_demo" -p 18271 root@142.170.89.112 -L 8080:localhost:8080

jsson QWEN:
{
  "models": [
    {
      "title": "Qwen 3 480B (Vast.ai)",
      "provider": "openai",
      "model": "qwen3-480b",
      "apiKey": "EMPTY",
      "baseUrl": "http://localhost:8000/v1",
      "contextLength": 32768,
      "completionOptions": {
        "temperature": 0.2,
        "maxTokens": 4096
      }
    }
  ],
  "tabAutocompleteModel": {
    "title": "Qwen 3 Autocomplete",
    "provider": "openai",
    "model": "qwen3-480b",
    "apiKey": "EMPTY",
    "baseUrl": "http://localhost:8000/v1"
  },
  "allowAnonymousTelemetry": false
}
a lancer dans l'ide
ssh -L 8000:localhost:8000 root@[IP_DE_L_INSTANCE] -p [PORT_SSH_VAST]