Models
Third-party models
Claude, GPT, Gemini, Grok, Llama and more — served directly, no API key required.
Served directly — no key required
Goatfied serves frontier third-party models directly. You don’t bring an API key and you don’t set up a provider account — they’re included with your plan and metered by tokens, exactly like our own GOAT models. Pick one in the model selector and go.
Available models
- Anthropic — Claude Opus 5, Claude Sonnet 5, Claude Haiku 5
- OpenAI — GPT-4.1, GPT-4.1 Mini
- xAI — Grok 4, Grok 3, Grok 3 Mini
- Google — Gemini 2.5 Pro, Gemini 2.5 Flash
- Meta — Llama 4 Maverick
- DeepSeek — DeepSeek R2, DeepSeek V3
- Mistral — Mistral Large
- Qwen — Qwen3 235B
- …plus a rotating line-up from Moonshot, Cohere, Amazon, Microsoft and more.
The picker is driven by a live catalog, so new provider releases appear day-one without waiting for an app update.
How billing works
Usage runs on two pools that reset each billing cycle. Our Auto + GOAT models draw from the cheaper Auto pool; every third-party model is charged at its published API rate from the API pool. When a pool’s included allowance is used up you can keep going on-demand at the same rates, or upgrade. See pricing and plans & usage for the included amounts per plan.
Switch models per task
Model preference is per-surface and per-task. Composer can default to Claude while Chat stays on Goat Frontier, and if a provider is rate-limited or down, requests fall back through an equivalent model so your work never stalls.
Bring your own key (optional)
Prefer to bill a provider directly? Add your own key under Settings → Models → Add provider. Keys are encrypted and never leave your machine (desktop) or your tenant (web), and requests on your key don’t draw from your Goatfied usage pools. For self-hosted or private models, point Goatfied at a custom endpoint.