Compare / Gemma 3 4B vs Mistral Small 3.2 24B

Gemma 3 4B vs Mistral Small 3.2 24B

Which is cheaper, which is more capable, and which does better at each kind of work, from 27 results on 4 sources that measured both models.

GoogleGemma 3 4B Released 11 Mar 2025 · $0.040 in · $0.050 out
Mistral AIMistral Small 3.2 24B Released 20 Jun 2025 · $0.094 in · $0.25 out
Gemma 3 4B wins 31 too close to callMistral Small 3.2 24B wins 23

Of the 27 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 16 November 2026 (catalogue release 2026-11-16-49bdfa045dee). See also Google vs Mistral AI, every model from both labs.

Gemma 3 4B or Mistral Small 3.2 24B? The short answer

Pick Mistral Small 3.2 24B if quality matters most: it scores higher (Intelligence Index 7.0 against 4.8). Pick Gemma 3 4B if cost matters more: it costs 3.1 times less per token.

  • Gemma 3 4B is cheaper: $0.042 against $0.13 per million tokens, 3.1 times less.
  • Mistral Small 3.2 24B scores higher on the Artificial Analysis Intelligence Index: 7.0 against 4.8.
  • Mistral Small 3.2 24B is newer: released 20 Jun 2025, against 11 Mar 2025 for Gemma 3 4B.
  • On third-party benchmarks, on the results both models have: Mistral Small 3.2 24B does better for agents and tool use, reasoning and knowledge and writing.

Head to head

4.8
Intelligence Index
7.0
$0.042
Price per million tokens, blendedlower is better
$0.13
131,072
Context window, tokens
256,000

Every fact side by side

Gemma 3 4BMistral Small 3.2 24B
Price per million tokens$0.040 in · $0.050 out$0.094 in · $0.25 out
Blended (3 in : 1 out)$0.042$0.13
Cheapest host, blended$0.042 (Parasail, BF16)$0.11 (DeepInfra, FP8)
Intelligence Index4.87.0
Spring Prompt overall––
Output speed, tokens a second––
Context window131,072 tokens256,000 tokens
Released11 Mar 202520 Jun 2025 (newer)
WeightsOpenOpen
Developer based inthe USFrance

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 158 of 159 119 of 159
Reasoning and knowledgeHard questions across science, maths and general knowledge 181 of 187 160 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality 157 of 169 123 of 169
Sticking to the factsSummarising without inventing things, and factual answers 32 of 139 not measured

From our research

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.

MeasureGemma 3 4BMistral Small 3.2 24B
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 255 · 182 of 387 measured 1,303 1,356
Business, management and finance · TextArena rating, higher is better 208 · 178 of 380 measured 1,330 1,363
Creative writing · TextArena rating, higher is better 243 · 182 of 385 measured 1,275 1,324
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 246 · 218 of 249 measured 4.8 7.0
GPQA Diamond% of questions, higher is better 227 · 202 of 232 measured 29.1% 50.5%
Humanity's Last Exam% of questions, higher is better 194 · 217 of 250 measured 5.3% 4.3%
Berkeley Function Calling Leaderboard (BFCL) V4 reported by Berkeley Function Calling Leaderboard (BFCL) V4 · 3 results
Overall accuracy · BFCL% correct, higher is better 74 · 38 of 81 measured 19.6%bfcl prompt 37.2%bfcl fc
Irrelevance detection · BFCL% correct, higher is better 71 · 15 of 81 measured 53.9%bfcl prompt 87.9%bfcl fc
Memory · BFCL% correct, higher is better 53 · 37 of 81 measured 8.6%bfcl prompt 18.1%bfcl fc
UGI Leaderboard reported by UGI Leaderboard · 3 results
Writing score · UGIscore out of 100, higher is better 185 · 122 of 230 measured 24.0 36.3
Requested-length error · UGI% off the requested word count, lower is better 33 · 83 of 230 measured 9.0% 17.0%
Style adherence · UGIscore from 0 to 1, higher is better 214 · 104 of 230 measured 0.29 0.35
Show all 27 results (15 more)
MeasureGemma 3 4BMistral Small 3.2 24B
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Expert prompts · TextArena rating, higher is better 270 · 205 of 338 measured 1,260 1,334
Instruction following · TextArena rating, higher is better 270 · 192 of 387 measured 1,267 1,338
Writing, literature and language · TextArena rating, higher is better 237 · 192 of 386 measured 1,298 1,329
Artificial Analysis reported by Artificial Analysis · 7 results
Terminal-Bench 2.1% of tasks, higher is better 127 · 110 of 131 measured 0.4% 5.6%
τ-bench banking% of tasks, higher is better 121 · 95 of 121 measured 0.4% 6.2%
Artificial Analysis Coding Indexindex score, higher is better 129 · 115 of 130 measured 2.7 12.5
IFBench% of instructions, higher is better 187 · 165 of 195 measured 28.3% 33.5%
Long-context reasoning (AA-LCR)% of questions, higher is better 226 · 209 of 241 measured 6.7% 20.3%
Terminal-Bench Hard% of tasks, higher is better 180 · 140 of 190 measured 0.8% 6.8%
τ²-bench telecom% of tasks, higher is better 187 · 136 of 193 measured 5.0% 29.5%
Berkeley Function Calling Leaderboard (BFCL) V4 reported by Berkeley Function Calling Leaderboard (BFCL) V4 · 5 results
Multi-turn tasks · BFCL% correct, higher is better 75 · 45 of 81 measured 0.4%bfcl prompt 14.8%bfcl prompt
Relevance detection · BFCL% correct, higher is better 1 · 8 of 81 measured 100.0%bfcl prompt 93.8%bfcl prompt
Single-turn calls (curated) · BFCL% correct, higher is better 74 · 4 of 81 measured 61.1%bfcl prompt 89.7%bfcl prompt
Single-turn calls (user-contributed) · BFCL% correct, higher is better 64 · 13 of 81 measured 60.8%bfcl prompt 79.1%bfcl prompt
Web search · BFCL% correct, higher is better 56 · 24 of 81 measured 1.0%bfcl prompt 31.0%bfcl fc

Quick answers

Is Gemma 3 4B better than Mistral Small 3.2 24B?

Pick Mistral Small 3.2 24B if quality matters most: it scores higher (Intelligence Index 7.0 against 4.8). Pick Gemma 3 4B if cost matters more: it costs 3.1 times less per token.

Which is cheaper, Gemma 3 4B or Mistral Small 3.2 24B?

Gemma 3 4B is cheaper: $0.042 against $0.13 per million tokens, 3.1 times less.

Which is better for agents and tool use, Gemma 3 4B or Mistral Small 3.2 24B?

Mistral Small 3.2 24B ranks 119 of 159 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 158 for Gemma 3 4B.

Which is better for reasoning and knowledge, Gemma 3 4B or Mistral Small 3.2 24B?

Mistral Small 3.2 24B ranks 160 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, Gemma 3 4B or Mistral Small 3.2 24B?

Mistral Small 3.2 24B ranks 123 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, Gemma 3 4B or Mistral Small 3.2 24B?

Mistral Small 3.2 24B: 256,000 tokens, against 131,072 for Gemma 3 4B.

Run Gemma 3 4B and Mistral Small 3.2 24B 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.