Gemma 3 27B vs Qwen2.5 VL 72B Instruct
Gemma 3 27B costs 3.3x less per output token than Qwen2.5 VL 72B Instruct ($0.30 vs $1.00 per million). Updated October 22, 2026; 5 benchmarks have a result for both.
Gemma 3 27B or Qwen2.5 VL 72B Instruct: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick Gemma 3 27B for output tokens 3.3x cheaper ($0.30 vs $1.00 per million) and a larger context window (131K against 128K tokens).
- Shared results
- 5
- Gemma 3 27B only
- 61
- Qwen2.5 VL 72B Instruct only
- 1
- Like-for-like categories
- 3 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
No call
Only Gemma 3 27B has a AA Intelligence Index result, 4.9.
Directional only
Agentic work
AA Agentic Index
No call
Only Gemma 3 27B has a AA Agentic Index result, 0.1.
Directional only
Long documents
Listed context window
Gemma 3 27B
Gemma 3 27B takes 131K tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Gemma 3 27B
Gemma 3 27B $0.0002, Qwen2.5 VL 72B Instruct $0.0013 per turn.
List prices
At a glance
- Coding index
- Gemma 3 27B not benchmarked, Qwen2.5 VL 72B Instruct not benchmarked.
- Widest category
- General, 10.4 points apart on average over 3 shared benchmarks.
- Widest single gap
- LMArena Vision: English: Gemma 3 27B 1188 against Qwen2.5 VL 72B Instruct 1115, 12.2 points apart on a 0 to 100 scale.
- Output price
- Gemma 3 27B $0.30 against Qwen2.5 VL 72B Instruct $1.00 per million output tokens, 3.3x apart.
- Context window
- Gemma 3 27B takes 131K tokens, Qwen2.5 VL 72B Instruct 128K.
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 27B | Bars | Qwen2.5 VL 72B Instruct | Ties |
|---|---|---|---|---|
| Reasoning | 0 | 1 | ||
| Knowledge | 0 | 1 | ||
| General | 3 | 0 | ||
| All shared boards | 3 | 2 |
CodingDirectional onlySix of these boards feed the Vibe Coding Index8 boards, none shared
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| Aider polyglot | 4.9% | Not measured | Only Gemma 3 27B measured |
| AA Coding Index | 10.1 | Not measured | Only Gemma 3 27B measured |
| LiveCodeBench (Artificial Analysis run) | 13.7% | Not measured | Only Gemma 3 27B measured |
| SciCode | 23.3% | Not measured | Only Gemma 3 27B measured |
| Terminal-Bench Hard (Artificial Analysis run) | 3.8% | Not measured | Only Gemma 3 27B measured |
| LMArena Text: coding | 1322 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: software and IT | 1354 | Not measured | Only Gemma 3 27B measured |
| SciCode | 0.0% | Not measured | Only Gemma 3 27B measured |
| Neither model has a result on the other 31 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| AA Intelligence Index | 4.9 | Not measured | Only Gemma 3 27B measured |
| Epoch Capabilities Index | 130.0 | Not measured | Only Gemma 3 27B measured |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| AA Agentic Index | 0.1 | Not measured | Only Gemma 3 27B measured |
| τ²-Bench Telecom | 10.5% | Not measured | Only Gemma 3 27B measured |
| OSWorld | Not measured | 5.0% | Only Qwen2.5 VL 72B Instruct measured |
| Berkeley Function Calling Leaderboard (BFCL) | 29.5% | Not measured | Only Gemma 3 27B measured |
| Neither model has a result on the other 37 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | 42.8% | Not measured | Only Gemma 3 27B measured |
| Humanity's Last Exam (Artificial Analysis run) | 4.4% | Not measured | Only Gemma 3 27B measured |
| GPQA Diamond (Epoch AI run) | 47.7% | Not measured | Only Gemma 3 27B measured |
| LMArena Text: hard prompts | 1340 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: expert | 1304 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: hard prompts in English | 1353 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision: diagrams | 1156 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision: homework | 1160 | Not measured | Only Gemma 3 27B measured |
| Kagi LLM Benchmark | 35.1% | 36.0% | Qwen2.5 VL 72B Instruct |
| Neither model has a result on the other 13 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| AA Math Index | 20.7 | Not measured | Only Gemma 3 27B measured |
| AIME 2025 (Artificial Analysis run) | 20.7% | Not measured | Only Gemma 3 27B measured |
| OTIS Mock AIME | 22.5% | Not measured | Only Gemma 3 27B measured |
| MATH level 5 | 74.0% | Not measured | Only Gemma 3 27B measured |
| LMArena Text: math | 1312 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: mathematical work | 1327 | Not measured | Only Gemma 3 27B measured |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | 66.9% | Not measured | Only Gemma 3 27B measured |
| GeoBench | 52.0% | 62.0% | Qwen2.5 VL 72B Instruct |
| LMArena Text: science | 1383 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: medicine and healthcare | 1391 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: legal and government | 1375 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: business and finance | 1360 | Not measured | Only Gemma 3 27B measured |
| FACTS Benchmark Suite | 29.1% | Not measured | Only Gemma 3 27B measured |
| Neither model has a result on the other 11 boards here | |||
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| Neither model has a result on 9 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | Leader |
|---|---|---|---|
| IFBench | 31.8% | Not measured | Only Gemma 3 27B measured |
| Long context reasoning (AA-LCR) | 7.3% | Not measured | Only Gemma 3 27B measured |
| DTBench | 52.5% | Not measured | Only Gemma 3 27B measured |
| LMCA | 12.3% | Not measured | Only Gemma 3 27B measured |
| Fiction.LiveBench | 33.3% | Not measured | Only Gemma 3 27B measured |
| Creative writing (Lech Mazur) | 79.9% | Not measured | Only Gemma 3 27B measured |
| LMArena Text | 1358 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: instruction following | 1321 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: creative writing | 1345 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision | 1165 | 1107 | Gemma 3 27B |
| LMArena Text: English | 1377 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: non-English | 1334 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: Chinese | 1346 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: German | 1362 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: Japanese | 1287 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: Korean | 1308 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: French | 1368 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: Spanish | 1349 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: Russian | 1349 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: Polish | 1345 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: multi-turn | 1345 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: longer queries | 1333 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: writing and language | 1347 | Not measured | Only Gemma 3 27B measured |
| LMArena Text: entertainment and media | 1320 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision: English | 1188 | 1115 | Gemma 3 27B |
| LMArena Vision: Chinese | 1164 | 1107 | Gemma 3 27B |
| LMArena Vision: reading text | 1156 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision: humor | 1166 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision: captioning | 1151 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision: recognizing things | 1196 | Not measured | Only Gemma 3 27B measured |
| LMArena Vision: creative writing | 1177 | Not measured | Only Gemma 3 27B measured |
| Neither model has a result on the other 15 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemma 3 27B | Qwen2.5 VL 72B Instruct | 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 27B costs less
Repository review
50K input + 3K output tokens
Gemma 3 27B costs less
Agent loop
200K input + 10K output tokens
Gemma 3 27B 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 27B | Qwen2.5 VL 72B Instruct |
|---|---|---|
| Maker | Qwen | |
| Weights | Open weights | Open weights |
| Released | March 12, 2025 | February 1, 2025 |
| Context window | 131K | 128K |
| Input $/Mtok | $0.08 | $0.80 |
| Output $/Mtok | $0.30 | $1.00 |
| 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 27B better than Qwen2.5 VL 72B Instruct for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Gemma 3 27B for output tokens 3.3x cheaper ($0.30 vs $1.00 per million) and a larger context window (131K against 128K tokens).
Which is cheaper, Gemma 3 27B or Qwen2.5 VL 72B Instruct?
Gemma 3 27B costs $0.08 per million input tokens and $0.30 per million output tokens; Qwen2.5 VL 72B Instruct costs $0.80 and $1.00. Cost per solved task is known for neither of them.
Which has the bigger context window, Gemma 3 27B or Qwen2.5 VL 72B Instruct?
Gemma 3 27B, with 131K tokens against 128K.
How many benchmarks is the Gemma 3 27B vs Qwen2.5 VL 72B Instruct comparison based on?
5 benchmarks have a result for both models as of October 22, 2026. Gemma 3 27B's coding index rests on 0 independent benchmarks and Qwen2.5 VL 72B Instruct's on 0. Where only one model has a result, the page says so instead of naming a winner.