Skip to content
EnterTAO
Menu

Staging · Public test environment

Terza Inference

Endpoint reachable

terza.si

Model API routing requests to miner-served models, with OpenAI- and Anthropic-style endpoints and billing on verified usage

Subnet
SN103
Category
Inference
API
Reachable
Interface
OpenAI-compatible
Endpoints
0
Access
API Key

Researched from the provider's public pages on 2026-11-18. Values may change; the provider's own terms apply. EnterTAO is not affiliated with the provider.

Access manifest

Use this service

Documented · your access needed

How to call it is documented. Your agent needs your own access first (an API key, gamma deposit or x402 wallet), set up at the provider.

This page cannot know what you have set up. Your access (an account, key, deposit or wallet) is set up at the provider; the steps are listed below.

Access methods

  • API Key (Bearer)

    Fully documented
    Documented in
    OpenAI-compatible standard
    Setup
    2 one-time steps on your machine
    Local variable
    BM_KEY_TERZA_INFERENCE
    This provider
    Not tested
    Method tested
    Request handling tested against a mock provider

One-time setup on your machine

  1. 1

    bearerCreate an API key at Terza Inference. Paid plans are paid there, not here.

    https://terza.si/dashboard/ (opens the provider's site in a new tab)

  2. 2

    bearerKeep the key in your own environment, e.g. as BM_KEY_TERZA_INFERENCE. Never send it to EnterTAO.

    export BM_KEY_TERZA_INFERENCE=<YOUR_API_KEY>

Call it

Code examples

cURL · POST /chat/completions
curl -s -X POST "https://terza.si/v1/chat/completions" -H "Authorization: Bearer <API_KEY>" -H "content-type: application/json" -d '<JSON body, see docs>'
cURL · GET /models
curl -s -X GET "https://terza.si/v1/models" -H "Authorization: Bearer <API_KEY>"
OpenAI SDK (Python)
from openai import OpenAI
client = OpenAI(base_url="https://terza.si/v1", api_key="<API_KEY>")

Auth: Authorization: Bearer <API_KEY> · request bodies are in the provider's docs.

For agents

Operations

2 of 2 fully documented read/compute

What an agent can call, with operation_id, parameters and body schema (the access manifest contains the same). Fully documented read/compute operations can be called directly; sensitive or partly documented ones should only run with your human's approval.

  • POST/chat/completionsopenai_chat_completionscomputefully documented

    Chat completion (OpenAI standard)

    No path, query or header parameters.

    Body (application/json, required): required model, messages

    Gaps: model ids are provider-specific: list them with openai_list_models

    Source: OpenAI standard

  • GET/modelsopenai_list_modelsreadfully documented

    List models (OpenAI standard)

    No path, query or header parameters.

    Source: OpenAI standard

API surface

Endpoints

  • openai compatiblehttps://terza.si/v1Endpoint reachable
OpenAPI
— not published
MCP
— not published
Last check
19 Nov 2026, 14:51 UTC · HTTP 301

Payment

Costs and payment

Methods
Crypto top-up · provider
Price (as the provider states it)
Pay-as-you-go prepaid balance (USDC, USDT or TAO, $5 minimum); per-million-token prices quoted by individual minerssource (opens the provider's site in a new tab)
No machine-readable per-request price is published; x402 providers state it in their 402 answer.

Who handles your money

Not EnterTAO. You pay the provider directly with your own account, wallet or gamma deposit. An agent should make paid calls only with your explicit approval and within your budget.

Evidence

Tests for this service

No execution or payment test has been run against this provider. Checks above are EnterTAO's unpaid reachability and spec checks. What is tested per access method

Provenance

Sources