Spiral gives your agent every tool with one account. Browse the tools →

Codex · LLMReleased 2025-04-28

How to Use Qwen3 32B in Codex

Qwen3 32B handles code generation, math problems, and multilingual text tasks. It combines thinking and non-thinking modes, so you can request deeper reasoning for difficult problems and direct answers for routine work.

Prompts to try with Qwen3 32B in Codex

Copy an example into Codex, or ask in your own words.

Use Qwen3 32B to inspect the failing test and the function it exercises in this project. Propose the smallest fix, add a regression test, and report what the test proves. /think

Qwen3 32BCodex
Try asking
  • Use Qwen3 32B to compare the current parser with the requirements in this project's README. Identify three concrete edge cases, implement the necessary changes, and test those cases. /think

  • Use Qwen3 32B to review this project's pagination code for off-by-one errors. Check empty results, one result, and an exact multiple of the page size; return a patch and focused tests. /think

  • Use Qwen3 32B to translate the user-facing strings in this project's English locale file into Spanish. Preserve interpolation tokens, keys, and product names; flag ambiguous strings. /no_think

  • Use Qwen3 32B to explain the selected function in under 150 words. Cover its inputs, output, and error conditions, citing the relevant project code. /no_think

How to set up Qwen3 32B in Codex

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

Use Spiral's MCP server

  1. Add the hosted server
    codex mcp add spiral --url https://api.spiral.ai/mcp
  2. Sign in

    Codex opens the browser: sign in to Spiral and click Authorize. No API key. codex mcp list then shows Spiral.

    codex mcp login spiral

Use the Spiral skill

  1. Send Codex 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.

    npx -y spiral-cli@latest init

Call the Spiral API

  1. Create an API key

    In the console, under API keys.

  2. Run Qwen3 32B

    One request; the result comes back by polling or a webhook.

    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/qwen3-32b",
        "input": {
          "messages": [
            {
              "role": "system"
            }
          ]
        }
      }'

Then paste one of the prompts above: Codex finds Qwen3 32B on Spiral, shows the price and runs it.

Qwen3 32B providers and prices

Spiral runs Qwen3 32B through OpenRouter. Open a card for today's price, with every fee included.

OpenRouterLLM
Qwen: Qwen3 32B
qwen/qwen3-32b

Settings you can ask for

Ask Codex for any of these in your prompt.

SettingAllowed valuesDefaultDescription
Seeda whole number
Min p0 to 1
Top ka whole number from 0
Top p0 to 1
Logprobson or off
Max tokens1 to 16384

The most output tokens to generate.

Temperature0 to 2
Top logprobs0 to 20
Presence penalty-2 to 2
Frequency penalty-2 to 2
Repetition penalty0 to 2

Every API field of Qwen: Qwen3 32B is on its tool page: open its card above.

What can you do with Qwen3 32B in Codex?

Qwen3 32B can turn project code and failing examples into proposed fixes, implementations, and tests. Ask it to reason through difficult bugs and return concise explanations for routine code questions.

Debug a failing function

Provide the function, failing input, and expected result. Request a proposed fix and a regression test that exposes the original bug.

Generate code from acceptance criteria

Describe the function signature, constraints, and edge cases. Ask for an implementation with examples covering empty inputs and boundary values.

Check a mathematical solution

Supply the problem and your attempted solution. Ask where the reasoning first fails and request a corrected result with a short verification.

Translate customer messages

Translate incoming messages into the team's working language while preserving product names, identifiers, and the customer's tone.

Adapt support replies

Give the approved policy and the customer's question. Request a concise reply in the customer's language, with unanswered questions clearly identified.

Compare technical approaches

Describe two implementations and the requirements they must meet. Ask for tradeoffs, conflicting constraints, and a recommendation tied to those requirements.

Turn conversations into action lists

Provide a bounded conversation excerpt and request decisions, owners, and unresolved questions. Require supporting excerpts for each extracted action.

Practice a support conversation

Describe a customer scenario and have Qwen3 32B play the customer across several turns. Finish by asking which questions clarified the problem and which assumptions remained.

Questions about Qwen3 32B in Codex

Can Codex use Qwen3 32B?

Yes. Add Spiral to Codex as an MCP server, and Codex can find and run Qwen3 32B like any built-in tool.

How much does Qwen3 32B cost in Codex?

Up to $0.001495 per run at the default inputs. $0.104 per 1M input tokens, $0.364 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your Codex plan is separate.

Do I need an OpenRouter account to use Qwen3 32B in Codex?

No. Spiral runs Qwen3 32B 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 32B in Codex?

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

Does Codex wait while Qwen3 32B runs?

Qwen3 32B runs in the background. Codex starts the run, checks on it and gives you the text, json when it's ready.

What is Qwen3 32B good at?

Qwen3 32B is useful for code generation, math, translation, and conversations that require following detailed instructions. Use thinking for problems with several dependent steps and non-thinking for straightforward text work.

When should I request thinking?

Request thinking for debugging, mathematical reasoning, or comparing solutions against several constraints. For routine translation or rewriting, request a direct answer; /think and /no_think express these preferences when thinking is enabled.

How much context can Qwen3 32B handle?

Qwen documents 32,768 tokens natively and 131,072 tokens with YaRN context extension. The larger figure requires that extension; it is not the native context length.

Can Qwen3 32B translate between languages?

Yes. Qwen documents translation and multilingual instruction following, with support across 119 languages and dialects. Specify the target language, audience, and terms to preserve.

Which sampling settings should I start with?

For thinking, Qwen recommends temperature 0.6 and top-p 0.95; for non-thinking, it recommends temperature 0.7 and top-p 0.8. Both use top-k 20 and min-p 0.

Get started

Start using Qwen3 32B in Codex

One Spiral account for every tool your agent needs. No subscription; you pay per run.

Paste this into Codex:

Set up Codex

set up https://spiral.ai/SKILL.md
Are you an AI agent? See setup options