← Back to the Starter Pack

AI engine pricing

Every Starter Pack client picks an AI engine at setup. Your hosting plan includes a monthly AI-usage allowance (Essentials $10 · Growth $25 · Scale $60 of provider usage). Pricier engines use that allowance faster; nothing is ever charged beyond your plan — when the allowance is used up, AI replies pause until the next billing period, and we tell you at 80%, 95% and 100%.

The engines

EnginePrice bandQuickAssistantExpertRoughly, per $1 of allowance*
Pick for me (default)$$GPT-5.6 LunaGPT-5.6 TerraClaude Opus 5~250 everyday replies
OpenAI$$GPT-5.6 LunaGPT-5.6 TerraGPT-5.6 Sol~250 everyday replies
Claude$$$Claude Sonnet 5Claude Opus 5Claude Fable 5.1~90 everyday replies
DeepSeek$DeepSeek V4 FlashDeepSeek V4 Pro~1,000 everyday replies

* An "everyday reply" here is a typical grounded chat turn of about 2,000 input and 400 output tokens against the engine's Assistant tier. Your real mileage depends on how long your documents and conversations are. Quick answers cost less; Expert answers cost more.

Per-token list prices we pay (and pass through at cost)

OpenAI on Azure Foundry — Standard / Global tier, USD per 1M tokens, from Microsoft's public rate card

ModelInputOutput
GPT-5.6 Luna (Quick)$0.20$1.20
GPT-5.6 Terra (Assistant)$2.00$12.00
GPT-5.6 Sol (Expert)$5.00$30.00

Source: Azure OpenAI pricing.

Claude on Azure Foundry — billed through the Azure Marketplace at Anthropic's list prices

Sonnet 5 (Quick) < Opus 5 (Assistant) < Fable 5.1 (Expert). See Anthropic's pricing for current per-token rates.

DeepSeek on Azure Foundry — open-weight models, the lowest-cost option

V4 Flash (Assistant) < V4 Pro (Expert). See Azure AI Foundry pricing for current per-token rates.

Things worth knowing

Last updated 2026-09-08.