SupaNexus
Get Started
OpenAI-compatible · DeepSeek Harness · 1M context

DeepSeek APIwithout the lock-in

Run DeepSeek V4 Pro and V4 Flash on SupaNexus — one Base URL, one key, OpenAI-compatible. Built for agents, including DeepSeek Harness custom providers.

Get a DeepSeek API keyDeepSeek Harness setup

Why SupaNexus

The DeepSeek endpoint your stack already speaks.

One Base URL for every model

DeepSeek today, Qwen or GLM tomorrow — same auth, same Chat Completions shape, same metering.

Ready for DeepSeek Harness

Add SupaNexus as a custom provider. Fetch the catalog over GET /v1/models and keep your agent runtime portable.

Zero retention, enterprise metering

Prompts are not stored for training. Usage, quotas, and spend caps sit in one console.

Catalog

DeepSeek models on SupaNexus

Use the Public API ID in Chat Completions, DeepSeek Harness, and any OpenAI-compatible client.

ModelPublic API IDContextBest for
DeepSeek V4 Prodeepseek/deepseek-v4-pro1M
ReasoningAgentsFlagship
DeepSeek V4 Flashdeepseek/deepseek-v4-flash1M
ChatThinkingValue

Live input / cached / output rates live in the console. See also homepage pricing · console catalog

Quickstart

First DeepSeek call in 30 seconds

Drop-in OpenAI SDK. Change base URL and API key. Use the Public API ID as model.

https://api.supanexus.ai/v1

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.supanexus.ai/v1",
    api_key="sk-snx-...",
)

response = client.chat.completions.create(
    model="deepseek/deepseek-v4-pro",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)

DeepSeek Harness

How to use DeepSeek Harness with a custom API

DeepSeek Harness is an open-source agent runtime. SupaNexus is not in the bundled provider catalog — add it as a custom provider and keep V4 Pro / V4 Flash on one key.

  1. 01

    Install DeepSeek Harness

    Follow the official developer preview at deepseek.com/harness. The product is iterating quickly; pin a version you have tested.

  2. 02

    Open Settings → Models

    Choose Add a custom provider. Catalog cards (DeepSeek official, OpenAI, Anthropic) will not point at SupaNexus.

  3. 03

    Set a permanent Provider ID

    Use a lowercase id such as supanexus. The id is stored on sessions and credentials — renaming later means creating a new provider.

  4. 04

    Base URL and protocol

    Base URL https://api.supanexus.ai/v1 (CN backup https://api.supanexus.io/v1). API protocol: OpenAI Chat Completions (openai-completions).

  5. 05

    Credential and models

    Paste a SupaNexus project API key. Add deepseek/deepseek-v4-pro and deepseek/deepseek-v4-flash, or use Fetch available models (GET /v1/models).

  6. 06

    Save, then optionally edit settings.yaml

    The UI stores the provider. For extra fields the form does not expose, open $DSH_HOME/settings.yaml and restart the harness.

Recommended settings.yaml snippet

Keep the key in an environment variable. Never commit it into the YAML file.

settings.yaml

llm-pi-ai:
  providers:
    supanexus:
      apiKeyEnv: SUPANEXUS_API_KEY
      api: openai-completions
      baseURL: https://api.supanexus.ai/v1
      models:
        - id: deepseek/deepseek-v4-pro
        - id: deepseek/deepseek-v4-flash

Recommended API

DeepSeek Harness recommended API on SupaNexus

Use this profile unless you already depend on Anthropic Messages elsewhere in the same agent.

Protocol

OpenAI Chat Completions (openai-completions). This is the native SupaNexus surface and the protocol Harness expects for a custom gateway.

Models

deepseek/deepseek-v4-pro for Harness / agent loops (flagship reasoning). deepseek/deepseek-v4-flash when latency and cost matter more than peak capability.

Why not the official DeepSeek card

The bundled DeepSeek provider talks to DeepSeek’s own API. SupaNexus gives you one key across vendors, unified metering, and no DeepSeek peak / off-peak split on our catalog.

Optional alternate protocol

If the rest of your tools already speak Anthropic, you can aim Harness at POST /v1/messages instead. For a greenfield Harness setup, stay on Chat Completions.

DeepSeek Harness (official)GitHub: deepseek-ai/deepseek-harnessFull API doc

FAQ

DeepSeek API questions

How teams actually wire DeepSeek on SupaNexus.

Use deepseek/deepseek-v4-pro for agent and hard reasoning work, and deepseek/deepseek-v4-flash when you want lower latency and cost. Do not send the legacy deepseek-chat alias.

Yes. Point the official OpenAI SDK at https://api.supanexus.ai/v1 with a SupaNexus project key. Streaming, tool calling, and JSON mode follow the Chat Completions contract.

Yes. Add a custom provider named supanexus, set the OpenAI Chat Completions protocol, and paste your project key. Details are on this page and in the API docs.

Global traffic uses api.supanexus.ai. The CN backup host is api.supanexus.io. Same API, same model IDs.

Other featured model APIs

Ship DeepSeek agents this afternoon.

Create a project, copy your key, and point DeepSeek Harness or any OpenAI SDK at SupaNexus.

Get a DeepSeek API keyDeepSeek Harness setup