NConnect
DocsDeepSeek V4 Pro + DeepSeek Harness

TUTORIAL

How to run DeepSeek V4 Pro with DeepSeek Harness

Run DeepSeek Harness (dsh) and its local web UI on DeepSeek V4 Pro, a 1M-context reasoning model, served by Nebius Token Factory through NConnect.

DeepSeek V4 ProDeepSeek Harness 5 minUpdated

DeepSeek Harness (dsh) is DeepSeek's open-source agent harness with a local web UI. NConnect layers Nebius in as a provider, so you can run DeepSeek V4 Pro, built for advanced reasoning, coding and long-horizon agent work, with a 1M-token context window.

Prerequisites

Steps

  1. Install NConnect

    This installs the CLI and the ndeepseek alias, plus Bun if needed.

    curl -fsSL https://nconnect.sh/install.sh | bash

  2. Add your Nebius API key

    nconnect configure

  3. Install DeepSeek Harness

    Skip this if dsh is already on your PATH. In an interactive terminal, NConnect also offers to run this for you the first time you launch DeepSeek Harness.

    npm install -g @deepseek-ai/dsh

  4. Launch DeepSeek Harness on DeepSeek V4 Pro

    nconnect --model deepseek-ai/DeepSeek-V4-Pro deepseek
    You should see
    NConnect ▸ Launching DeepSeek Harness with Nebius Token Factory (DeepSeek V4 Pro). Alpha.

    dsh boots its web profile by default, which serves the local web UI. Pass --profile headless to answer a single task and exit instead.

What NConnect does on launch

  • Writes a config patch that adds Nebius as a provider and pins the model. The patch holds no credentials; the key is passed by environment variable.
  • Runs dsh against a throwaway home. dsh remembers the last model picked in its web UI, and that setting would otherwise override --model. Your real ~/.dsh is untouched.

Track cost and add fallback

DeepSeek Harness is a spawned harness, so by default it calls Nebius directly and NConnect reports $0.00. Route it through the daemon to meter every turn and get automatic fallback if DeepSeek V4 Pro is busy:

NCONNECT_METER=1 nconnect --model deepseek-ai/DeepSeek-V4-Pro deepseek

Then check spend with nconnect usage --last 7d. See Cost metering.

More tutorials

Kimi K3Claude CodeHow to run Kimi K3 with Claude CodeUse Claude Code with Moonshot AI's Kimi K3 and its full 1M-token context, served by Nebius Token Factory through NConnect. No Anthropic account needed. Video included. 5 minRead tutorial GLM 5.3Codex CLIHow to run GLM 5.3 with Codex CLIRun OpenAI's Codex CLI on Z.ai's GLM 5.3, a 1M-context model for coding and tool use, served by Nebius Token Factory through NConnect. No OpenAI account needed. 5 minRead tutorial Qwen 3.5 397BOpenCodeHow to run Qwen 3.5 with OpenCodeRun the OpenCode terminal agent on Alibaba's Qwen 3.5 397B, served by Nebius Token Factory through NConnect, without editing your OpenCode config. 5 minRead tutorial MiniMax M3Pi CodeHow to run MiniMax M3 with Pi CodeRun the Pi coding agent on MiniMax M3, a fast and low-cost reasoning model, served by Nebius Token Factory through NConnect with a temporary provider config. 5 minRead tutorial