Plenty of tools pick one model and hide it from you. We took the opposite bet: let people use the AI they trust, switch when it makes sense, and default to a smart router when they'd rather not think about it. No single model wins every task, and the landscape changes monthly — so flexibility isn't a feature, it's a hedge against lock-in.
Four providers, one workspace
Gate 15 runs on Claude, GLM, GPT, and Gemini, with a curated list of the models actually worth using for serious coding. You can set a default for your account or override it per turn from the composer — switch models mid-conversation without losing context.
Auto mode, when you'd rather not choose
Most of the time you don't want to think about model selection — you want a good answer. Auto mode routes each turn to a strong default so you get reliable results without managing it yourself. When you do have a preference, it's one click away.
Bring your own keys
On paid plans you can add your own Anthropic, OpenAI, Google, or Z.ai API keys. Usage then bills through those providers directly, and your keys stay encrypted at rest — never displayed back to you in full, never exposed to the agent.
- Pick Claude, GLM, GPT, or Gemini per turn or per account.
- Auto mode routes to a strong default when you don't want to decide.
- Tune reasoning effort — from low up to max — to trade speed for depth.
- Add your own provider keys; they stay encrypted and out of the model's view.
The goal is simple: you should never feel stuck with a model that's wrong for the job. The best AI for a task is the one you get to choose.