[Integrations](https://spiral.ai/integrations) › [OpenClaw](https://spiral.ai/integrations/openclaw) › [Granite 4.2 8B](https://spiral.ai/integrations/openclaw/granite-4-2-8b)

# How to Use Granite 4.2 8B in OpenClaw

Granite 4.2 8B handles coding, tool use, and multi-step reasoning. Unlike Granite 4.2 3B, it received additional agent training for software engineering, terminal tasks, and search.

|  |  |
| --- | --- |
| Model | Granite 4.2 8B |
| Available through | OpenRouter |
| Category | LLM |
| Price | Up to $0.000803 per run at the default inputs. $0.078 per 1M input tokens, $0.195 per 1M output tokens. |
| Returns | text, json |
| Runs | In the background |
| Setup | About 2 minutes, no API key |

## Prompts to try with Granite 4.2 8B in OpenClaw

- Each morning, use Granite 4.2 8B to review new incident reports available through connected tools. Group reports by symptom, quote the evidence for each suspected cause, and list the next diagnostic checks.
- Each weekday, use Granite 4.2 8B to research changes announced on the official product pages in the watch list. Report confirmed changes with source links and distinguish announcements from speculation.
- Each evening, use Granite 4.2 8B to draft replies for new Spanish support messages available through connected tools. Follow the approved policy and Spanish examples, preserve product names, and queue the drafts for review.
- Every Monday, use Granite 4.2 8B to compare the current policy text available through connected tools with the previous version. Report changed obligations, supporting passages, and unresolved contradictions.
- Every hour, use Granite 4.2 8B to compare new service alerts with available status-tool results. Report confirmed incidents separately from unverified alerts and identify any missing evidence.

## How to set up Granite 4.2 8B in OpenClaw

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

### Use the Spiral skill

1. **Send OpenClaw 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 Spiral's MCP server

1. **Add the server**: On the machine that runs OpenClaw's Gateway:

```bash
openclaw mcp add spiral \
  --url https://api.spiral.ai/mcp \
  --transport streamable-http \
  --auth oauth
```

2. **Sign in**: In Settings › MCP, select Sign in on the Spiral server, sign in to Spiral and click Authorize. No API key.

```bash
openclaw mcp login spiral
```

3. **Reload and check**: A running Gateway may keep its old tool list: restart it, or Save & Publish in the Control UI.

```bash
openclaw mcp reload
openclaw mcp doctor spiral --probe
```

### 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 Granite 4.2 8B**: 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": "ibm-granite/granite-4.2-8b",
    "input": {
      "messages": [
        {
          "role": "system"
        }
      ]
    }
  }'
```

Then paste one of the prompts above: OpenClaw finds Granite 4.2 8B on Spiral, shows the price and runs it.

Setup guide in the docs: https://spiral.ai/docs/quickstarts/openclaw

## Granite 4.2 8B tools and prices

- [IBM: Granite 4.2 8B](https://spiral.ai/tools/openrouter/ibm-granite/granite-4.2-8b) (OpenRouter): Up to $0.000803 per run at the default inputs. $0.078 per 1M input tokens, $0.195 per 1M output tokens.

## Settings you can ask for

| Setting | Allowed values | Default | Description |
| --- | --- | --- | --- |
| seed |  | — |  |
| stop |  | — |  |
| min_p | 0 to 1 | — |  |
| tools |  | — |  |
| top_k | 0 to … | — |  |
| top_p | 0 to 1 | — |  |

## What can you do with Granite 4.2 8B in OpenClaw?

Review recurring operational questions and prepare evidence-backed reports with Granite 4.2 8B. Define the sources, decision rules, and report format so each scheduled run produces a result you can check.

### Fix a reproducible software bug

Ask Granite 4.2 8B to trace a failing test, propose a focused patch, and check the result against that test.

### Diagnose a terminal failure

Give Granite 4.2 8B command output and the expected result. Request the next diagnostic command and an explanation of what its result would establish.

### Research a question across sources

Have Granite 4.2 8B use available search tools to collect evidence, compare conflicting claims, and draft an answer with citations.

### Answer questions about policy text

Supply relevant policy passages to Granite 4.2 8B and request an answer with supporting excerpts and unresolved questions.

### Draft multilingual support replies

Give Granite 4.2 8B an approved policy, a customer message, and examples in a supported language. Request a reply that preserves product names and policy terms.

### Investigate a service incident

Ask Granite 4.2 8B to compare incident reports with available status-tool results, separate observations from hypotheses, and identify the next check.

### Plan a multi-step tool task

Give Granite 4.2 8B a goal and clearly described tools. Ask it to identify dependencies, request missing information, and verify tool results before continuing.

## Overview

- [OpenRouter](https://spiral.ai/providers/openrouter)
- [All OpenClaw guides](https://spiral.ai/integrations/openclaw)

## Granite 4.2 8B in other agents

- [Amp](https://spiral.ai/integrations/amp/granite-4-2-8b)
- [ChatGPT](https://spiral.ai/integrations/chatgpt/granite-4-2-8b)
- [Claude](https://spiral.ai/integrations/claude-desktop/granite-4-2-8b)
- [Claude Code](https://spiral.ai/integrations/claude-code/granite-4-2-8b)
- [Cline](https://spiral.ai/integrations/cline/granite-4-2-8b)
- [Codex](https://spiral.ai/integrations/codex/granite-4-2-8b)
- [Cursor](https://spiral.ai/integrations/cursor/granite-4-2-8b)
- [Devin](https://spiral.ai/integrations/devin/granite-4-2-8b)
- [dots](https://spiral.ai/integrations/dots/granite-4-2-8b)
- [Factory Droid](https://spiral.ai/integrations/factory-droid/granite-4-2-8b)
- [Gemini](https://spiral.ai/integrations/gemini/granite-4-2-8b)
- [Gemini CLI](https://spiral.ai/integrations/gemini-cli/granite-4-2-8b)
- [Google Antigravity](https://spiral.ai/integrations/antigravity/granite-4-2-8b)
- [goose](https://spiral.ai/integrations/goose/granite-4-2-8b)
- [Grok](https://spiral.ai/integrations/grok/granite-4-2-8b)
- [Grok Build](https://spiral.ai/integrations/grok-build/granite-4-2-8b)
- [Hermes](https://spiral.ai/integrations/hermes/granite-4-2-8b)
- [Instinct](https://spiral.ai/integrations/instinct/granite-4-2-8b)
- [Manus](https://spiral.ai/integrations/manus/granite-4-2-8b)
- [Muse](https://spiral.ai/integrations/muse/granite-4-2-8b)
- [Muse Code](https://spiral.ai/integrations/muse-code/granite-4-2-8b)
- [n8n](https://spiral.ai/integrations/n8n/granite-4-2-8b)
- [Notion](https://spiral.ai/integrations/notion/granite-4-2-8b)
- [opencode](https://spiral.ai/integrations/opencode/granite-4-2-8b)
- [Perplexity](https://spiral.ai/integrations/perplexity/granite-4-2-8b)
- [Raycast](https://spiral.ai/integrations/raycast/granite-4-2-8b)
- [VS Code](https://spiral.ai/integrations/vscode/granite-4-2-8b)
- [Windsurf](https://spiral.ai/integrations/windsurf/granite-4-2-8b)
- [Zapier](https://spiral.ai/integrations/zapier/granite-4-2-8b)

## More OpenClaw guides

- [Gemma 4 31B](https://spiral.ai/integrations/openclaw/gemma-4-31b)
- [Grok 4.20](https://spiral.ai/integrations/openclaw/grok-4-20)
- [Gemma 4 26B A4B](https://spiral.ai/integrations/openclaw/gemma-4-26b-a4b)
- [Grok 4.20 Multi-Agent](https://spiral.ai/integrations/openclaw/grok-4-20-multi-agent)
- [Gemma 3 4B](https://spiral.ai/integrations/openclaw/gemma-3-4b)
- [Grok 4.3](https://spiral.ai/integrations/openclaw/grok-4-3)

## Questions about Granite 4.2 8B in OpenClaw

**Can OpenClaw use Granite 4.2 8B?**

Yes. Add Spiral to OpenClaw as an MCP server, and OpenClaw can find and run Granite 4.2 8B like any built-in tool.

**How much does Granite 4.2 8B cost in OpenClaw?**

Up to $0.000803 per run at the default inputs. $0.078 per 1M input tokens, $0.195 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your OpenClaw plan is separate.

**Do I need an OpenRouter account to use Granite 4.2 8B in OpenClaw?**

No. Spiral runs Granite 4.2 8B on its own OpenRouter account. You sign in to Spiral once and never paste a key.

**How long does it take to set up Granite 4.2 8B in OpenClaw?**

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

**Does OpenClaw wait while Granite 4.2 8B runs?**

Granite 4.2 8B runs in the background. OpenClaw starts the run, checks on it and gives you the text, json when it's ready.

**What is Granite 4.2 8B useful for?**

Use it for coding tasks, tool-assisted work, and problems that require several decisions. IBM specifically trained it on software engineering, terminal coding, and search workflows.

**How much context can Granite 4.2 8B handle?**

IBM documents a native 128K-token context window. For long-document work, ask a focused question and request supporting passages so you can check the answer.

**Does Granite 4.2 always need to think before answering?**

It supports thinking, non-thinking, and low-effort thinking modes. Use full thinking for multi-step problems, brief thinking for simpler decisions, and direct answers for straightforward requests.

**Which languages does Granite 4.2 support?**

IBM lists English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, and Chinese. Performance can differ from English, so include examples of the wording and format you want.

**Which sampling settings should I start with?**

IBM recommends temperature 1.0 and top-p 0.95 for chat, reasoning, and tool calling. Start there before experimenting with other values.

