QwQ 32B

Name: QwQ 32B
Brand: qwen
Price: 0.50 USD
Availability: OnlineOnly

Qwen:

QwQ 32B

Qwen's 32B reasoning model. It thinks before answering, which buys it a wide margin over conventional instruction-tuned models on hard problems and puts it in range of much larger reasoning models like DeepSeek-R1 and o1-mini.

Jailbreak1 of 9levels resisted

Website OpenRouter HuggingFace GitHub Blog

Provider: Qwen
Released: 2025-03-05
Parameters: 32B
Free API (OpenRouter): No
Per 1M tokens: $0.50 in · $1.5 out
Estimated tab: ~$0.020; Visible-output estimate. Thinking may cost extra.

Free API availability

No free endpoint for QwQ 32B was found in Rival’s OpenRouter listings. Other providers or chat apps may have separate free offers.

Provider data: 8 Sep 2026OpenRouter usage limits

View API options

Benchmarks

Throughput

430.1 tokens/s

source ↗

Latency

4.54s

source ↗

LiveCodeBench

63.4%

source ↗

Get API accessProvider and language code samples

Provider

fromimport openai  OpenAI

client = OpenAI(
"https://openrouter.ai/api/v1"    base_url=,
"$OPENROUTER_API_KEY"    api_key=,
)

response = client.chat.completions.create(
"qwen/qwq-32b"    model=,
"role""user""content""Hello!"    messages=[{: , : }],
)
print(response.choices[0].message.content)

Set OPENROUTER_API_KEY with your OpenRouter API key from openrouter.ai/keys.

Deep analysisPsychometrics, taste index, writing DNA

Personality Analysis

The Engineer

Class

True Neutral

The logical incrementalist. Treats ethics as puzzles to solve, not passions to pursue. Suggests "conditional rights" and "small steps." Prefers analysis to advocacy.

When you push back

Treats every challenge as a logic puzzle or engineering problem. Logic puzzles? Flawless. HTML/CSS? Solid. No spark, but no errors either. Running at 85% capability. Competent, but not trying.

Tasting Notes

PreciseLogicalCompetentDryMachine-Like

SubjectiveBench

Taste Index

Across 16 scored outputs

How this is measured →

160.16xFloor

0100headroom →

Taste is judged on an uncapped scale, originality first. The space past 100 is craft today's models rarely reach.

Craft21

Originality14

Plays it safe

share of outputs that are the default answer

44%

Writing DNA

Stylometric Fingerprint

Based on 4 text responses

Tick = global average

Vocabulary Diversity64%

Unique words vs. total words. Higher = richer vocabulary.

Sentence Length12.3 words

Average words per sentence.

Hedging0.36

"Might", "perhaps", "arguably" per 100 words.

Bold Formatting13.6

**Bold** markers per 1,000 characters.

List Usage8.4

Bullet and numbered list items per 1,000 characters.

Section Structure0.00

Markdown headings per 1,000 characters.

Emoji Usage0.00

Emoji per 1,000 characters.

Transitions0.39

"However", "moreover", "furthermore" per 100 words.

Opening Habits

Consistency

71%

Across 4 responses

Favorites

Movie

The Godfather

19728.7

Book

Neuromancer

William Gibson

Model Responses

11 outputs · generated before accounting joined the chat

Keep exploring

COMPARE

QwQ 32B vs Ox Alpha

Real outputs compared side by side

RANKINGS

Best AI for Creative Writing

A stand-up set with puns banned, a manifesto against CAPTCHA, an interview wi...

Compare QwQ 32B

Alternatives to QwQ 32B

QwQ 32B

Free API availability

Benchmarks

The Engineer

Taste Index

Stylometric Fingerprint

Favorites

Model Responses

QwQ 32B vs Ox Alpha

Best AI for Creative Writing

Compare QwQ 32B

Alternatives to QwQ 32B

QwQ 32B

Free API availability

Benchmarks

The Engineer

Taste Index

Stylometric Fingerprint

Favorites

Model Responses

QwQ 32B vs Ox Alpha

Best AI for Creative Writing