Compare / Qwen3-32B vs gpt-oss-120b

Qwen3-32B vs gpt-oss-120b

Which is cheaper, which is more capable, and which does better at each kind of work, from 23 results on 4 sources that measured both models.

AlibabaQwen3-32B Released 28 Apr 2025 · $0.14 in · $0.40 out
OpenAIgpt-oss-120b Released 5 Aug 2025 · $0.15 in · $0.60 out
Qwen3-32B wins 48 too close to callgpt-oss-120b wins 11

Of the 23 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 9 October 2026 (catalogue release 2026-10-09-dd620e778ac5). See also Alibaba vs OpenAI, every model from both labs.

Qwen3-32B or gpt-oss-120b? The short answer

Pick gpt-oss-120b if quality matters most: it scores higher (Intelligence Index 11.6 against 8.6). Pick Qwen3-32B if cost matters more: it costs 22% less per token.

  • Qwen3-32B is cheaper: $0.21 against $0.26 per million tokens, 22% less.
  • gpt-oss-120b scores higher on the Artificial Analysis Intelligence Index: 11.6 against 8.6.
  • gpt-oss-120b is newer: released 5 Aug 2025, against 28 Apr 2025 for Qwen3-32B.
  • On third-party benchmarks, on the results both models have: Qwen3-32B does better for sticking to the facts; gpt-oss-120b for agents and tool use, reasoning and knowledge and writing.

Head to head

8.6
Intelligence Index
11.6
$0.21
Price per million tokens, blendedlower is better
$0.26
131,072
Context window, tokens
131,072

Every fact side by side

Qwen3-32Bgpt-oss-120b
Price per million tokens$0.14 in · $0.40 out$0.15 in · $0.60 out
Blended (3 in : 1 out)$0.21$0.26
Cheapest host, blended$0.13 (DeepInfra, FP8)$0.065 (CoreWeave, FP4)
Intelligence Index8.611.6
Spring Prompt overall––
Output speed, tokens a second–188
Context window131,072 tokens131,072 tokens
Released28 Apr 20255 Aug 2025 (newer)
WeightsOpenOpen
Developer based inChinathe US

Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.

Which is better for what

Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.

Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 113 of 164 120 of 164
Professional workReal tasks from banking, consulting and law, business documents and freelance projects not measured 49 of 51
Reasoning and knowledgeHard questions across science, maths and general knowledge 148 of 198 114 of 198
WritingWhat people prefer in blind comparisons, and judged writing quality 139 of 188 133 of 188
Sticking to the factsSummarising without inventing things, and factual answers 22 of 152 139 of 152

Every shared result

Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.

MeasureQwen3-32Bgpt-oss-120b
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 195 · 187 of 387 measured 1,347 1,352
Business, management and finance · TextArena rating, higher is better 203 · 183 of 380 measured 1,340 1,352
Creative writing · TextArena rating, higher is better 205 · 238 of 385 measured 1,304 1,278
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 188 · 157 of 243 measured 8.6thinking reasoning 11.6high reasoning
GPQA Diamond% of questions, higher is better 167 · 118 of 231 measured 66.8%thinking reasoning 78.2%high reasoning
Humanity's Last Exam% of questions, higher is better 168 · 110 of 242 measured 7.4%thinking reasoning 19.6%high reasoning
UGI Leaderboard reported by UGI Leaderboard · 3 results
Writing score · UGIscore out of 100, higher is better 134 · 116 of 230 measured 33.0no reasoning 38.5medium reasoning
Requested-length error · UGI% off the requested word count, lower is better 99 · 224 of 230 measured 19.0%no reasoning 107.0%medium reasoning
Style adherence · UGIscore from 0 to 1, higher is better 175 · 217 of 230 measured 0.32no reasoning 0.29medium reasoning
Vectara Hallucination Leaderboard reported by Vectara Hallucination Leaderboard · 2 results
Hallucination rate · Vectara% of summaries, lower is better 24 · 88 of 104 measured 5.9% 14.2%
Answer rate · Vectara% of documents, higher is better 17 · 17 of 104 measured 99.9% 99.9%
Show all 23 results (12 more)
MeasureQwen3-32Bgpt-oss-120b
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Expert prompts · TextArena rating, higher is better 159 · 188 of 338 measured 1,398 1,357
Instruction following · TextArena rating, higher is better 198 · 204 of 387 measured 1,331 1,323
Writing, literature and language · TextArena rating, higher is better 217 · 215 of 386 measured 1,308 1,309
Artificial Analysis reported by Artificial Analysis · 9 results
Terminal-Bench 2.1% of tasks, higher is better 111 · 87 of 130 measured 5.2%thinking reasoning 26.2%high reasoning
Terminal-Bench 4.0% of tasks, higher is better 66 · 66 of 112 measured 0.0%thinking reasoning 0.0%high reasoning
τ-bench banking% of tasks, higher is better 102 · 67 of 121 measured 5.4%thinking reasoning 12.8%high reasoning
Artificial Analysis Coding Indexindex score, higher is better 111 · 86 of 131 measured 15.3thinking reasoning 30.4high reasoning
IFBench% of instructions, higher is better 155 · 58 of 195 measured 36.3%thinking reasoning 69.0%high reasoning
Long-context reasoning (AA-LCR)% of questions, higher is better 222 · 152 of 233 measured 0.0%no reasoning 52.0%high reasoning
SciCode% of problems, higher is better 91 · 93 of 113 measured 36.0%thinking reasoning 34.0%high reasoning
Terminal-Bench Hard% of tasks, higher is better 169 · 84 of 190 measured 3.0%thinking reasoning 23.5%high reasoning
τ²-bench telecom% of tasks, higher is better 135 · 86 of 193 measured 29.8%thinking reasoning 65.8%high reasoning

Quick answers

Is Qwen3-32B better than gpt-oss-120b?

Pick gpt-oss-120b if quality matters most: it scores higher (Intelligence Index 11.6 against 8.6). Pick Qwen3-32B if cost matters more: it costs 22% less per token.

Which is cheaper, Qwen3-32B or gpt-oss-120b?

Qwen3-32B is cheaper: $0.21 against $0.26 per million tokens, 22% less.

Which is better for agents and tool use, Qwen3-32B or gpt-oss-120b?

Qwen3-32B ranks 113 of 164 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 120 for gpt-oss-120b.

Which is better for reasoning and knowledge, Qwen3-32B or gpt-oss-120b?

gpt-oss-120b ranks 114 of 198 models on sale for hard questions across science, maths and general knowledge, against 148 for Qwen3-32B.

Which is better for writing, Qwen3-32B or gpt-oss-120b?

gpt-oss-120b ranks 133 of 188 models on sale for what people prefer in blind comparisons, and judged writing quality, against 139 for Qwen3-32B.

Which is better for sticking to the facts, Qwen3-32B or gpt-oss-120b?

Qwen3-32B ranks 22 of 152 models on sale for summarising without inventing things, and factual answers, against 139 for gpt-oss-120b.

Run Qwen3-32B and gpt-oss-120b on your own prompt

Benchmarks aren't your data. Try both side by side in the playground, with the cost of every answer.

Try the playgroundTalk to us