Compare / Qwen2.5-72B-Instruct vs Gemma 3 4B

Qwen2.5-72B-Instruct vs Gemma 3 4B

Which is cheaper, which is more capable, and which does better at each kind of work, from 15 results on 3 sources that measured both models.

AlibabaQwen2.5-72B-Instruct Released 17 Sep 2024 · $0.36 in · $0.40 out
GoogleGemma 3 4B Released 11 Mar 2025 · $0.040 in · $0.050 out
Qwen2.5-72B-Instruct wins 85 too close to callGemma 3 4B wins 2

Of the 15 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 19 November 2026 (catalogue release 2026-11-19-3608133e8331). See also Alibaba vs Google, every model from both labs.

Qwen2.5-72B-Instruct or Gemma 3 4B? The short answer

Pick Qwen2.5-72B-Instruct if quality matters most: it scores higher (Intelligence Index 7.7 against 4.8). Pick Gemma 3 4B if cost matters more: it costs 8.7 times less per token.

  • Gemma 3 4B is cheaper: $0.042 against $0.37 per million tokens, 8.7 times less.
  • Qwen2.5-72B-Instruct scores higher on the Artificial Analysis Intelligence Index: 7.7 against 4.8.
  • Gemma 3 4B is newer: released 11 Mar 2025, against 17 Sep 2024 for Qwen2.5-72B-Instruct.
  • On third-party benchmarks, on the results both models have: Qwen2.5-72B-Instruct does better for reasoning and knowledge and writing.

Head to head

7.7
Intelligence Index
4.8
$0.37
Price per million tokens, blendedlower is better
$0.042
32,768
Context window, tokens
131,072

Every fact side by side

Qwen2.5-72B-InstructGemma 3 4B
Price per million tokens$0.36 in · $0.40 out$0.040 in · $0.050 out
Blended (3 in : 1 out)$0.37$0.042
Cheapest host, blended–$0.042 (Parasail, BF16)
Intelligence Index7.74.8
Spring Prompt overall––
Output speed, tokens a second––
Context window32,768 tokens131,072 tokens
Released17 Sep 202411 Mar 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 not measured 158 of 159
Reasoning and knowledgeHard questions across science, maths and general knowledge 167 of 187 181 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality 146 of 169 157 of 169
Sticking to the factsSummarising without inventing things, and factual answers not measured 32 of 139

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.

MeasureQwen2.5-72B-InstructGemma 3 4B
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 257 · 255 of 387 measured 1,303 1,303
Business, management and finance · TextArena rating, higher is better 245 · 208 of 380 measured 1,299 1,330
Creative writing · TextArena rating, higher is better 265 · 243 of 385 measured 1,254 1,275
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 205 · 246 of 249 measured 7.7 4.8
GPQA Diamond% of questions, higher is better 207 · 227 of 232 measured 49.1% 29.1%
Humanity's Last Exam% of questions, higher is better 237 · 194 of 250 measured 3.6% 5.3%
UGI Leaderboard reported by UGI Leaderboard · 3 results
Writing score · UGIscore out of 100, higher is better 147 · 185 of 230 measured 31.5 24.0
Requested-length error · UGI% off the requested word count, lower is better 124 · 33 of 230 measured 21.0% 9.0%
Style adherence · UGIscore from 0 to 1, higher is better 64 · 214 of 230 measured 0.36 0.29
Show all 15 results (6 more)
MeasureQwen2.5-72B-InstructGemma 3 4B
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Expert prompts · TextArena rating, higher is better 244 · 270 of 338 measured 1,295 1,260
Instruction following · TextArena rating, higher is better 245 · 270 of 387 measured 1,292 1,267
Writing, literature and language · TextArena rating, higher is better 253 · 237 of 386 measured 1,282 1,298
Artificial Analysis reported by Artificial Analysis · 3 results
IFBench% of instructions, higher is better 150 · 187 of 195 measured 36.9% 28.3%
Terminal-Bench Hard% of tasks, higher is better 156 · 180 of 190 measured 4.5% 0.8%
τ²-bench telecom% of tasks, higher is better 123 · 187 of 193 measured 34.5% 5.0%

Quick answers

Is Qwen2.5-72B-Instruct better than Gemma 3 4B?

Pick Qwen2.5-72B-Instruct if quality matters most: it scores higher (Intelligence Index 7.7 against 4.8). Pick Gemma 3 4B if cost matters more: it costs 8.7 times less per token.

Which is cheaper, Qwen2.5-72B-Instruct or Gemma 3 4B?

Gemma 3 4B is cheaper: $0.042 against $0.37 per million tokens, 8.7 times less.

Which is better for reasoning and knowledge, Qwen2.5-72B-Instruct or Gemma 3 4B?

Qwen2.5-72B-Instruct ranks 167 of 187 models on sale for hard questions across science, maths and general knowledge, against 181 for Gemma 3 4B.

Which is better for writing, Qwen2.5-72B-Instruct or Gemma 3 4B?

Qwen2.5-72B-Instruct ranks 146 of 169 models on sale for what people prefer in blind comparisons, and judged writing quality, against 157 for Gemma 3 4B.

Which has the bigger context window, Qwen2.5-72B-Instruct or Gemma 3 4B?

Gemma 3 4B: 131,072 tokens, against 32,768 for Qwen2.5-72B-Instruct.

Run Qwen2.5-72B-Instruct and Gemma 3 4B on your own prompt.

Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.