Gemma 3 12B vs Qwen3 32B
Qwen3 32B leads Gemma 3 12B on the Artificial Analysis Intelligence Index, 7.3 to 3.8, and Gemma 3 12B costs 2.5x less per output token ($0.1625 vs $0.40 per million). Updated October 21, 2026; 40 benchmarks have a result for both.
Gemma 3 12B or Qwen3 32B: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick Gemma 3 12B for output tokens 2.5x cheaper ($0.1625 vs $0.40 per million) and work measured by LiveCodeBench (Artificial Analysis run) or SciCode, where it leads.
- Shared results
- 40
- Gemma 3 12B only
- 3
- Qwen3 32B only
- 5
- Like-for-like categories
- 7 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Neither has a coding score yet.
One side not measured
Intelligence
AA Intelligence Index
Qwen3 32B
Qwen3 32B leads, 7.3 against 3.8.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Gemma 3 12B has a AA Agentic Index result, 0.1.
Directional only
Long documents
Listed context window
No call
Both take 131K tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Gemma 3 12B
Gemma 3 12B $0.0002, Qwen3 32B $0.0003 per turn.
List prices
At a glance
- Coding index
- Gemma 3 12B not benchmarked, Qwen3 32B not benchmarked.
- Widest category
- Math, 37.8 points apart on average over 5 shared benchmarks.
- Widest single gap
- AA Math Index: Gemma 3 12B 18.3 against Qwen3 32B 73.0, 54.7 points apart on a 0 to 100 scale.
- Output price
- Gemma 3 12B $0.1625 against Qwen3 32B $0.40 per million output tokens, 2.5x apart.
- Context window
- Both take 131K tokens.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | Gemma 3 12B | Bars | Qwen3 32B | Ties |
|---|---|---|---|---|
| Coding | 0 | 5 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 2 | ||
| Reasoning | 0 | 6 | ||
| Math | 0 | 5 | ||
| Knowledge | 2 | 3 | ||
| General | 9 | 6 | ||
| All shared boards | 11 | 29 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index05
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| Aider polyglot | Not measured | 40.0% | Only Qwen3 32B measured |
| AA Coding Index | 5.8 | Not measured | Only Gemma 3 12B measured |
| LiveCodeBench (Artificial Analysis run) | 13.7% | 54.6% | Qwen3 32B |
| SciCode | 16.4% | 36.0% | Qwen3 32B |
| Terminal-Bench Hard (Artificial Analysis run) | 0.8% | 3.0% | Qwen3 32B |
| LMArena Text: coding | 1281 | 1357 | Qwen3 32B |
| LMArena Text: software and IT | 1313 | 1366 | Qwen3 32B |
| SciCode | 0.0% | Not measured | Only Gemma 3 12B measured |
| Neither model has a result on the other 31 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| AA Intelligence Index | 3.8 | 7.3 | Qwen3 32B |
| Epoch Capabilities Index | 123.5 | 138.4 | Qwen3 32B |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| AA Agentic Index | 0.1 | Not measured | Only Gemma 3 12B measured |
| τ²-Bench Telecom | 10.8% | 29.8% | Qwen3 32B |
| Berkeley Function Calling Leaderboard (BFCL) | 30.4% | 48.7% | Qwen3 32B |
| BrowseComp-Plus | Not measured | 10.7% | Only Qwen3 32B measured |
| Neither model has a result on the other 37 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | 34.9% | 66.8% | Qwen3 32B |
| Humanity's Last Exam (Artificial Analysis run) | 4.2% | 7.4% | Qwen3 32B |
| GPQA Diamond (Epoch AI run) | 39.5% | 65.7% | Qwen3 32B |
| LMArena Text: hard prompts | 1309 | 1334 | Qwen3 32B |
| LMArena Text: expert | 1248 | 1362 | Qwen3 32B |
| LMArena Text: hard prompts in English | 1318 | 1350 | Qwen3 32B |
| Kagi LLM Benchmark | Not measured | 48.7% | Only Qwen3 32B measured |
| Neither model has a result on the other 15 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index05
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| AA Math Index | 18.3 | 73.0 | Qwen3 32B |
| AIME 2025 (Artificial Analysis run) | 18.3% | 73.0% | Qwen3 32B |
| OTIS Mock AIME | 16.7% | 66.9% | Qwen3 32B |
| LMArena Text: math | 1307 | 1399 | Qwen3 32B |
| LMArena Text: mathematical work | 1336 | 1420 | Qwen3 32B |
| Neither model has a result on the other 8 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index23
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | 59.5% | 79.8% | Qwen3 32B |
| LMArena Text: science | 1366 | 1372 | Qwen3 32B |
| LMArena Text: medicine and healthcare | 1334 | 1368 | Qwen3 32B |
| LMArena Text: legal and government | 1381 | 1324 | Gemma 3 12B |
| LMArena Text: business and finance | 1376 | 1318 | Gemma 3 12B |
| Neither model has a result on the other 13 boards here | |||
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| Neither model has a result on 9 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index96
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| IFBench | 36.7% | 36.3% | Gemma 3 12B |
| Long context reasoning (AA-LCR) | 8.3% | 0.0% | Gemma 3 12B |
| DTBench | 48.8% | 67.5% | Qwen3 32B |
| LMCA | 4.5% | 17.3% | Qwen3 32B |
| Fiction.LiveBench | Not measured | 74.2% | Only Qwen3 32B measured |
| LMArena Text | 1334 | 1340 | Qwen3 32B |
| LMArena Text: instruction following | 1299 | 1305 | Qwen3 32B |
| LMArena Text: creative writing | 1331 | 1297 | Gemma 3 12B |
| LMArena Text: English | 1349 | 1359 | Qwen3 32B |
| LMArena Text: non-English | 1318 | 1317 | Gemma 3 12B |
| LMArena Text: Chinese | Not measured | 1357 | Only Qwen3 32B measured |
| LMArena Text: German | 1370 | 1341 | Gemma 3 12B |
| LMArena Text: Russian | 1335 | 1311 | Gemma 3 12B |
| LMArena Text: multi-turn | 1334 | 1331 | Gemma 3 12B |
| LMArena Text: longer queries | 1317 | 1327 | Qwen3 32B |
| LMArena Text: writing and language | 1307 | 1298 | Gemma 3 12B |
| LMArena Text: entertainment and media | 1297 | 1279 | Gemma 3 12B |
| Neither model has a result on the other 29 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemma 3 12B | Qwen3 32B | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Gemma 3 12B costs less
Repository review
50K input + 3K output tokens
Gemma 3 12B costs less
Agent loop
200K input + 10K output tokens
Gemma 3 12B costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Google Gemma 3 12B | Qwen3 32B |
|---|---|---|
| Maker | Qwen | |
| Weights | Open weights | Open weights |
| Released | March 13, 2025 | April 28, 2025 |
| Context window | 131K | 131K |
| Input $/Mtok | $0.075 | $0.14 |
| Output $/Mtok | $0.1625 | $0.40 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Gemma 3 12B better than Qwen3 32B for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Gemma 3 12B for output tokens 2.5x cheaper ($0.1625 vs $0.40 per million) and work measured by LiveCodeBench (Artificial Analysis run) or SciCode, where it leads.
Which is cheaper, Gemma 3 12B or Qwen3 32B?
Gemma 3 12B costs $0.075 per million input tokens and $0.1625 per million output tokens; Qwen3 32B costs $0.14 and $0.40. Cost per solved task is known for neither of them.
Which has the bigger context window, Gemma 3 12B or Qwen3 32B?
Neither: both take 131K tokens.
How many benchmarks is the Gemma 3 12B vs Qwen3 32B comparison based on?
40 benchmarks have a result for both models as of October 21, 2026. Gemma 3 12B's coding index rests on 0 independent benchmarks and Qwen3 32B's on 0. Where only one model has a result, the page says so instead of naming a winner.