The Hoshi 1.0 API.
Frontier agent intelligence, a drop-in away.
Call Hoshi 1.0 like any OpenAI completion endpoint — change three lines and keep your stack. Per-token pricing, model-agnostic and open, with no vendor lock-in.
Pay per token
Hoshi 1.0 API pricing
Metered per token — no subscription required for raw API access.
Input
Tokens you send in — prompts, context, and tool definitions.
Cached input
Repeated context served from cache, ~15x cheaper than fresh input.
Output
Tokens Hoshi generates — completions, code, and structured output.
Prices reflect the published Hoshi 1.0 token rates. The Shogo agent plans (Free/Basic/Pro/Business) are billed separately on the pricing page.
A drop-in for OpenAI
Already using the OpenAI SDK? Point the base URL at https://api.shogo.ai/v1, set your key, and use hoshi-1.0.
curl https://api.shogo.ai/v1/chat/completions \
-H "Authorization: Bearer $SHOGO_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "hoshi-1.0",
"messages": [
{ "role": "user", "content": "Refactor this function for readability." }
]
}'Built to slot in
OpenAI-compatible
Drop-in Chat Completions API. Point your existing OpenAI client at our base URL and set the model — no rewrite.
Model-agnostic & open
Hoshi 1.0 joins a routing table alongside Anthropic, OpenAI, Google, and local models. Pick by score, cost, or privacy.
No vendor lock-in
Standard request/response shape and OpenAI SDKs. Switch back or route around us anytime — nothing proprietary to unwind.
Frontier-tier evals
Hoshi 1.0 sweeps independent agentic, structured-output, and coding evals — at roughly a tenth of the cost.
FAQ
Yes. The API speaks the standard Chat Completions shape, so the official OpenAI SDKs and most OpenAI-compatible tooling work unchanged. You only swap three things: the base URL, your API key, and the model id ("hoshi-1.0").
The Shogo subscription plans (Free, Basic, Pro, Business) cover the self-evolving agents that build software you own. The Hoshi API is the raw model endpoint, metered per token — input, cached input, and output — for developers who want to call Hoshi directly from their own code.
Hoshi 1.0 today. Because Shogo is model-agnostic and open, the same endpoint sits in front of a routing table that also includes Anthropic, OpenAI, Google, and local models — so you can route by score, cost, or privacy per request.
Request API access from your Shogo workspace. We'll provision a key you can drop straight into your existing OpenAI client.