[Integrations](https://spiral.ai/integrations) › [Muse Code](https://spiral.ai/integrations/muse-code) › [Qwen3 235B A22B Thinking 2507](https://spiral.ai/integrations/muse-code/qwen3-235b-a22b-thinking-2507)

# How to Use Qwen3 235B A22B Thinking 2507 in Muse Code

Qwen3 235B A22B Thinking 2507 suits difficult math, code, and logic problems. Unlike Qwen3 235B A22B Instruct 2507, it always uses thinking mode and is designed for deeper reasoning.

|  |  |
| --- | --- |
| Model | Qwen3 235B A22B Thinking 2507 |
| Available through | OpenRouter |
| Category | LLM |
| Price | Up to $0.018658 per run at the default inputs. $0.585 per 1M input tokens, $4.55 per 1M output tokens. |
| Returns | text, json |
| Runs | In the background |
| Setup | About 2 minutes, no API key |

## Prompts to try with Qwen3 235B A22B Thinking 2507 in Muse Code

- Use Qwen3 235B A22B Thinking 2507 to investigate the first failing test in this project. Trace the relevant functions, identify the cause, and implement the smallest fix with a regression test.
- Use Qwen3 235B A22B Thinking 2507 to review this project's retry logic. Check duplicate requests, timeouts, and exhausted retries, then propose tests for any uncovered behavior.
- Use Qwen3 235B A22B Thinking 2507 to compare the implementation with the project's specification. Report contradictory requirements and code mismatches with file references.
- Use Qwen3 235B A22B Thinking 2507 to find the slowest algorithm in the supplied profiling results. Compare two replacements, explain complexity, and implement the better-supported choice.
- Use Qwen3 235B A22B Thinking 2507 to check the Spanish translations in this repository against the English source. Flag changed conditions, quantities, and technical terms.

## How to set up Qwen3 235B A22B Thinking 2507 in Muse Code

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

### Use Spiral's MCP server

1. **Install Muse Code**

```bash
curl -fsSL https://dev.meta.ai/install.sh | bash
```

2. **Ask Muse to add the server**: Start `muse` and ask:

```text
Add the Spiral MCP server at https://api.spiral.ai/mcp
```

3. **Sign in**: In your terminal, run:

```bash
muse mcp login spiral
```

### Use the Spiral skill

1. **Send Muse Code 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 Qwen3 235B A22B Thinking 2507**: 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": "qwen/qwen3-235b-a22b-thinking-2507",
    "input": {
      "messages": [
        {
          "role": "system"
        }
      ]
    }
  }'
```

Then paste one of the prompts above: Muse Code finds Qwen3 235B A22B Thinking 2507 on Spiral, shows the price and runs it.

Setup guide in the docs: https://spiral.ai/docs/quickstarts/muse-code

## Qwen3 235B A22B Thinking 2507 tools and prices

- [Qwen: Qwen3 235B A22B Thinking 2507](https://spiral.ai/tools/openrouter/qwen/qwen3-235b-a22b-thinking-2507) (OpenRouter): Up to $0.018658 per run at the default inputs. $0.585 per 1M input tokens, $4.55 per 1M output tokens.

## Settings you can ask for

| Setting | Allowed values | Default | Description |
| --- | --- | --- | --- |
| stop |  | — |  |
| tools |  | — |  |
| top_k | 0 to … | — |  |
| top_p | 0 to 1 | — |  |
| reasoning |  | — | Reasoning effort; reasoning tokens count toward max_tokens. |
| max_tokens | 1 to 16384 | — | The most output tokens to generate. |

## What can you do with Qwen3 235B A22B Thinking 2507 in Muse Code?

Turn difficult code failures into a proposed fix with Qwen3 235B A22B Thinking. Ask your coding agent to connect the relevant project files, compare causes, and verify the result with targeted tests.

### Debug a failure across several functions

Trace a failing test through the relevant code, compare possible causes, and propose a small patch with a regression check.

### Design an algorithm under constraints

Turn requirements into an algorithm, explain its time and memory costs, and test boundary cases before implementation.

### Check a mathematical solution

Find the first unsupported step in a proposed solution, correct it, and verify the result with substitution or a second method.

### Compare scientific explanations

Analyze supplied observations, identify assumptions behind competing explanations, and suggest a test that could distinguish them.

### Find contradictions in long specifications

Compare requirements across supplied text, flag incompatible rules, and produce a short list of decisions needed before implementation.

### Review multilingual technical content

Compare a translation with its source, checking whether conditions, quantities, and technical terms preserve the original meaning.

### Plan a task that uses tools

Choose which available tool can resolve each missing fact, then use the returned evidence to prepare a recommendation.

## Overview

- [OpenRouter](https://spiral.ai/providers/openrouter)
- [All Muse Code guides](https://spiral.ai/integrations/muse-code)

## Qwen3 235B A22B Thinking 2507 in other agents

- [Amp](https://spiral.ai/integrations/amp/qwen3-235b-a22b-thinking-2507)
- [ChatGPT](https://spiral.ai/integrations/chatgpt/qwen3-235b-a22b-thinking-2507)
- [Claude](https://spiral.ai/integrations/claude-desktop/qwen3-235b-a22b-thinking-2507)
- [Claude Code](https://spiral.ai/integrations/claude-code/qwen3-235b-a22b-thinking-2507)
- [Cline](https://spiral.ai/integrations/cline/qwen3-235b-a22b-thinking-2507)
- [Codex](https://spiral.ai/integrations/codex/qwen3-235b-a22b-thinking-2507)
- [Cursor](https://spiral.ai/integrations/cursor/qwen3-235b-a22b-thinking-2507)
- [Devin](https://spiral.ai/integrations/devin/qwen3-235b-a22b-thinking-2507)
- [dots](https://spiral.ai/integrations/dots/qwen3-235b-a22b-thinking-2507)
- [Factory Droid](https://spiral.ai/integrations/factory-droid/qwen3-235b-a22b-thinking-2507)
- [Gemini](https://spiral.ai/integrations/gemini/qwen3-235b-a22b-thinking-2507)
- [Gemini CLI](https://spiral.ai/integrations/gemini-cli/qwen3-235b-a22b-thinking-2507)
- [Google Antigravity](https://spiral.ai/integrations/antigravity/qwen3-235b-a22b-thinking-2507)
- [goose](https://spiral.ai/integrations/goose/qwen3-235b-a22b-thinking-2507)
- [Grok](https://spiral.ai/integrations/grok/qwen3-235b-a22b-thinking-2507)
- [Grok Build](https://spiral.ai/integrations/grok-build/qwen3-235b-a22b-thinking-2507)
- [Hermes](https://spiral.ai/integrations/hermes/qwen3-235b-a22b-thinking-2507)
- [Instinct](https://spiral.ai/integrations/instinct/qwen3-235b-a22b-thinking-2507)
- [Manus](https://spiral.ai/integrations/manus/qwen3-235b-a22b-thinking-2507)
- [Muse](https://spiral.ai/integrations/muse/qwen3-235b-a22b-thinking-2507)
- [n8n](https://spiral.ai/integrations/n8n/qwen3-235b-a22b-thinking-2507)
- [Notion](https://spiral.ai/integrations/notion/qwen3-235b-a22b-thinking-2507)
- [OpenClaw](https://spiral.ai/integrations/openclaw/qwen3-235b-a22b-thinking-2507)
- [opencode](https://spiral.ai/integrations/opencode/qwen3-235b-a22b-thinking-2507)
- [Perplexity](https://spiral.ai/integrations/perplexity/qwen3-235b-a22b-thinking-2507)
- [Raycast](https://spiral.ai/integrations/raycast/qwen3-235b-a22b-thinking-2507)
- [VS Code](https://spiral.ai/integrations/vscode/qwen3-235b-a22b-thinking-2507)
- [Windsurf](https://spiral.ai/integrations/windsurf/qwen3-235b-a22b-thinking-2507)
- [Zapier](https://spiral.ai/integrations/zapier/qwen3-235b-a22b-thinking-2507)

## More Muse Code guides

- [Qwen3 235B A22B Instruct 2507](https://spiral.ai/integrations/muse-code/qwen3-235b-a22b-instruct-2507)
- [Qwen3 30B A3B](https://spiral.ai/integrations/muse-code/qwen3-30b-a3b)
- [Qwen3 14B](https://spiral.ai/integrations/muse-code/qwen3-14b)
- [Qwen3 30B A3B Instruct 2507](https://spiral.ai/integrations/muse-code/qwen3-30b-a3b-instruct-2507)
- [Qwen2.5-VL 72B Instruct](https://spiral.ai/integrations/muse-code/qwen2-5-vl-72b-instruct)
- [Qwen3 32B](https://spiral.ai/integrations/muse-code/qwen3-32b)

## Questions about Qwen3 235B A22B Thinking 2507 in Muse Code

**Can Muse Code use Qwen3 235B A22B Thinking 2507?**

Yes. Add Spiral to Muse Code as an MCP server, and Muse Code can find and run Qwen3 235B A22B Thinking 2507 like any built-in tool.

**How much does Qwen3 235B A22B Thinking 2507 cost in Muse Code?**

Up to $0.018658 per run at the default inputs. $0.585 per 1M input tokens, $4.55 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your Muse Code plan is separate.

**Do I need an OpenRouter account to use Qwen3 235B A22B Thinking 2507 in Muse Code?**

No. Spiral runs Qwen3 235B A22B Thinking 2507 on its own OpenRouter account. You sign in to Spiral once and never paste a key.

**How long does it take to set up Qwen3 235B A22B Thinking 2507 in Muse Code?**

About two minutes: install Muse Code, then ask Muse to add the server, then sign in.

**Does Muse Code wait while Qwen3 235B A22B Thinking 2507 runs?**

Qwen3 235B A22B Thinking 2507 runs in the background. Muse Code starts the run, checks on it and gives you the text, json when it's ready.

**What is Qwen3 235B A22B Thinking best at?**

Qwen recommends it for complex reasoning, including math, science, coding, and logic. Try tasks with several constraints, competing explanations, or an answer you can check.

**Can I turn thinking off?**

Qwen3 235B A22B Thinking 2507 only supports thinking mode. Qwen3 235B A22B Instruct 2507 is the separately trained non-thinking model.

**How much text can it consider?**

The maker lists a native context window of 262,144 tokens. That describes context capacity, not the length of a single reply.

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

Qwen recommends temperature 0.6, top-p 0.95, and top-k 20. Use those as a starting point when comparing results on your own tasks.

**Can it work in languages other than English?**

Qwen documents 119 languages and dialects for Qwen3, including Chinese, Spanish, Arabic, and Japanese. Specify the answer language and provide a glossary when terminology matters.

