Endearo · settings

Models & providers

Endearo can answer with several AI models. DeepSeek V4 Pro and V4 Flash are the defaults; turning on advanced settings adds Claude Opus 4.8, Grok 4.3, and Gemini. You can also save your own API keys and run a side-by-side comparison →.

TierProviderModelFormatPricing
V4 ProDeepSeekdeepseek-v4-proOpenAIsee DeepSeek pricing
V4 FlashDeepSeekdeepseek-v4-flashOpenAIsee DeepSeek pricing
Claude Opus 4.8 advancedAnthropicclaude-opus-4-8Anthropicsee Anthropic pricing
Grok 4.3 advancedxAIgrok-4.3OpenAI$1.25 in / $2.50 out per 1M · 1M ctx
Gemini 3.1 Pro advancedGooglegemini-3.1-proOpenAIsee Gemini pricing
Gemini 3.5 Flash advancedGooglegemini-3.5-flashOpenAIsee Gemini pricing

Claude Fable 5 and Claude Mythos 5 are suspended (Anthropic, 2026-06-12) and are never offered — use Opus 4.8 or another tier.

Two API formats, one path

DeepSeek speaks both the OpenAI format (api.deepseek.com) and an Anthropic-compatible one (api.deepseek.com/anthropic); Claude speaks the Anthropic format and xAI/Grok the OpenAI format. Endearo auto-switches on the tier — one Anthropic-format caller serves both Claude and DeepSeek-via-/anthropic.

If a model is unavailable

  • DeepSeek out of balance → “the website will come back soon.”
  • An advanced model out of balance / bad key → it explains, offers “save your own key”, and falls back to V4 Pro so the chat still answers.
  • Multiple keys per provider auto-swap seamlessly (primary → backup).

Your own keys

Save your own provider keys in Settings (used for both the comparison page and the chat, deletable anytime). They’re sent per-request to your Endearo agent — never to anyone else.

Keeping pricing + models current

Prices change, so this page links the live source of truth rather than hardcoding numbers, and endearo models --latest checks the configured model ids against DeepSeek’s live API.

— Endearo · model settings · back to the assistant