[Integrations](https://spiral.ai/integrations) › [Gemini CLI](https://spiral.ai/integrations/gemini-cli) › [Hy4](https://spiral.ai/integrations/gemini-cli/hy4)

# How to Use Hy4 in Gemini CLI

Hy4 is suited to complex coding and research tasks that require sustained planning. Compared with Hy4 Preview, it improves long-task execution and supports up to 65,536 output tokens.

|  |  |
| --- | --- |
| Model | Hy4 |
| Available through | OpenRouter |
| Category | LLM |
| Price | Up to $0.006679 per run at the default inputs. $0.5421 per 1M input tokens, $1.62565 per 1M output tokens. |
| Returns | text, json |
| Runs | In the background |
| Setup | About 2 minutes, no API key |

## Prompts to try with Hy4 in Gemini CLI

- Use Hy4 to investigate the failing tests in this project. Trace the cause, implement the smallest justified fix, and run the affected tests.
- Use Hy4 to implement the settings form described in the project requirements. Include validation, saving, and error handling, then check each requirement.
- Ask Hy4 through Spiral to review the diff on this branch and list the three riskiest changes. Save the result in ./assets.
- Get a second opinion from Hy4 on this function, then tell me what the call cost.
- Summarize these release notes with Hy4 in five bullet points.

## How to set up Hy4 in Gemini CLI

Pick the way that suits you. Each takes about two minutes.

### Use Spiral's MCP server

1. **Add the server**: Or add it to `~/.gemini/settings.json` yourself:

```bash
gemini mcp add --scope user --transport http spiral https://api.spiral.ai/mcp
```

2. **Sign in**: Start `gemini` and run:

```text
/mcp auth spiral
```

3. **Check it**: In Gemini CLI, run `/mcp`. From your shell:

```bash
gemini mcp list
```

### Use the Spiral skill

1. **Send Gemini CLI this message**: Your agent reads Spiral's skill and sets itself up.

```
set up https://spiral.ai/SKILL.md
```

2. **Sign in or paste a key**: Approve Spiral in the browser, or give the agent a key from the console.

### Use the Spiral CLI

1. **Install and sign in**: Installs the Spiral CLI, signs you in and adds its skill to your coding agents.

```bash
npx -y spiral-cli@latest init
```

### Call the Spiral API

1. **Create an API key**: In the console, under API keys.

2. **Run Hy4**: One request; the result comes back by polling or a webhook.

```bash
curl https://api.spiral.ai/v1/runs \
  -H "Authorization: Bearer $SPIRAL_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
    "provider": "openrouter",
    "model": "tencent/hy4",
    "input": {
      "messages": [
        {
          "role": "system"
        }
      ]
    }
  }'
```

Then paste one of the prompts above: Gemini CLI finds Hy4 on Spiral, shows the price and runs it.

Setup guide in the docs: https://spiral.ai/docs/quickstarts/gemini-cli

## Hy4 tools and prices

- [Tencent: Hy4](https://spiral.ai/tools/openrouter/tencent/hy4) (OpenRouter): Up to $0.006679 per run at the default inputs. $0.5421 per 1M input tokens, $1.62565 per 1M output tokens.

## Settings you can ask for

| Setting | Allowed values | Default | Description |
| --- | --- | --- | --- |
| stop |  | — |  |
| tools |  | — |  |
| reasoning |  | — | Reasoning effort; reasoning tokens count toward max_tokens. |
| max_tokens | 1 to 65536 | — | The most output tokens to generate. |
| temperature | 0 to 2 | — |  |
| tool_choice |  | — |  |

## What can you do with Hy4 in Gemini CLI?

Tencent Hy4 can help carry code changes from investigation through implementation and testing. Give it the relevant project context and concrete acceptance checks.

### Fix a bug across several files

Trace a failing behavior through the relevant code, propose a focused patch, and check it against the reported failure.

### Plan a code migration

Compare existing code with the target requirements, identify dependencies, and break the migration into changes with acceptance checks.

### Turn working notes into a report

Combine supplied notes into a report with decisions, unresolved questions, and evidence attached to each conclusion.

### Review a financial model

Examine supplied assumptions and tables, explain the calculations, and flag inconsistencies for review.

### Prototype a small game

Build a playable loop from a written brief, then refine controls, rules, and failure states through testing.

### Develop a research computation

Translate a mathematical or scientific question into a computational approach with explicit assumptions and validation checks.

## Overview

- [OpenRouter](https://spiral.ai/providers/openrouter)
- [All Gemini CLI guides](https://spiral.ai/integrations/gemini-cli)

## Hy4 in other agents

- [Amp](https://spiral.ai/integrations/amp/hy4)
- [ChatGPT](https://spiral.ai/integrations/chatgpt/hy4)
- [Claude](https://spiral.ai/integrations/claude-desktop/hy4)
- [Claude Code](https://spiral.ai/integrations/claude-code/hy4)
- [Cline](https://spiral.ai/integrations/cline/hy4)
- [Codex](https://spiral.ai/integrations/codex/hy4)
- [Cursor](https://spiral.ai/integrations/cursor/hy4)
- [Devin](https://spiral.ai/integrations/devin/hy4)
- [dots](https://spiral.ai/integrations/dots/hy4)
- [Factory Droid](https://spiral.ai/integrations/factory-droid/hy4)
- [Gemini](https://spiral.ai/integrations/gemini/hy4)
- [Google Antigravity](https://spiral.ai/integrations/antigravity/hy4)
- [goose](https://spiral.ai/integrations/goose/hy4)
- [Grok](https://spiral.ai/integrations/grok/hy4)
- [Grok Build](https://spiral.ai/integrations/grok-build/hy4)
- [Hermes](https://spiral.ai/integrations/hermes/hy4)
- [Instinct](https://spiral.ai/integrations/instinct/hy4)
- [Manus](https://spiral.ai/integrations/manus/hy4)
- [Muse](https://spiral.ai/integrations/muse/hy4)
- [Muse Code](https://spiral.ai/integrations/muse-code/hy4)
- [n8n](https://spiral.ai/integrations/n8n/hy4)
- [Notion](https://spiral.ai/integrations/notion/hy4)
- [OpenClaw](https://spiral.ai/integrations/openclaw/hy4)
- [opencode](https://spiral.ai/integrations/opencode/hy4)
- [Perplexity](https://spiral.ai/integrations/perplexity/hy4)
- [Raycast](https://spiral.ai/integrations/raycast/hy4)
- [VS Code](https://spiral.ai/integrations/vscode/hy4)
- [Windsurf](https://spiral.ai/integrations/windsurf/hy4)
- [Zapier](https://spiral.ai/integrations/zapier/hy4)

## More Gemini CLI guides

- [Hy3](https://spiral.ai/integrations/gemini-cli/hy3)
- [Hy4 Flash](https://spiral.ai/integrations/gemini-cli/hy4-flash)
- [Hy-MT2-7B](https://spiral.ai/integrations/gemini-cli/hy-mt2-7b)
- [Hy4 Preview](https://spiral.ai/integrations/gemini-cli/hy4-preview)
- [Hy-MT2-30B-A3B](https://spiral.ai/integrations/gemini-cli/hy-mt2-30b-a3b)
- [Inkling](https://spiral.ai/integrations/gemini-cli/inkling)

## Questions about Hy4 in Gemini CLI

**Can Gemini CLI use Hy4?**

Yes. Add Spiral to Gemini CLI as an MCP server, and Gemini CLI can find and run Hy4 like any built-in tool.

**How much does Hy4 cost in Gemini CLI?**

Up to $0.006679 per run at the default inputs. $0.5421 per 1M input tokens, $1.62565 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your Gemini CLI plan is separate.

**Do I need an OpenRouter account to use Hy4 in Gemini CLI?**

No. Spiral runs Hy4 on its own OpenRouter account. You sign in to Spiral once and never paste a key.

**How long does it take to set up Hy4 in Gemini CLI?**

About two minutes: add the server, then sign in, then check it.

**Does Gemini CLI wait while Hy4 runs?**

Hy4 runs in the background. Gemini CLI starts the run, checks on it and gives you the text, json when it's ready.

**What is Tencent Hy4 useful for?**

Coding, office analysis, and scientific research are documented uses. Give it work that requires connecting evidence and completing several related steps.

**How much context can Tencent Hy4 handle?**

Hy4 and Hy4 Preview each have a 1,048,576-token context window. For a large request, identify the relevant sections and state the exact question you need answered.

**How long can a Tencent Hy4 response be?**

Hy4 supports up to 65,536 completion tokens; Hy4 Preview supports up to 64,000. Specify the deliverable's length instead of requesting the maximum by default.

**Can Tencent Hy4 make factual mistakes?**

Yes. Tencent reported lower hallucination rates for Hy4 than Hy4 Preview in an internal evaluation, but errors remained. Ask for supporting passages and verify consequential claims against the original evidence.

**Can Tencent Hy4 help build a game?**

Game prototyping is a documented use. Start with a small playable loop, define the controls and win condition, and refine it after testing.

