Model comparison
DeepSeek-V3.2-Exp vs Gemma 3 27B
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 30.5 on the Noometry Index. Gemma 3 27B costs 2.9× less per token, which makes it the better buy when DeepSeek-V3.2-Exp's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
DeepSeek
GoogleSummary
- They share 31 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 9 categories and Gemma 3 27B in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3.2-Exp leads 51.6 to 25.4.
- The biggest single-benchmark swing is Aider Polyglot: 74.2% for DeepSeek-V3.2-Exp and 4.9% for Gemma 3 27B.
- Gemma 3 27B is cheaper at $0.08 / $0.16 per million input/output tokens, against $0.26 / $0.38 for DeepSeek-V3.2-Exp.
- DeepSeek-V3.2-Exp accepts more context: 164K tokens versus 131K.
Side by side
| DeepSeek-V3.2-Exp | Gemma 3 27B | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 44.9 | 30.5 |
| Released | 2025-09-29 | 2025-03-11 |
| Weights | Open | Open |
| Context window | 164K | 131K |
| Max output | 66K | 8K |
| Input $ / M tokens | $0.26 | $0.08 |
| Output $ / M tokens | $0.38 | $0.16 |
| Results tracked | 48 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 46.4 (#61), Gemma 3 27B: 22.1 (#329)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| Aider Polyglot | 74.2% | 4.9% |
| SciCode | 38.9% | 21.2% |
| LMArena Coding | 1454 | 1322 |
| SWE-bench Verified (bash only) | 70% | — |
| LMArena WebDev | 1362 | — |
| SWE-bench Multilingual | 59% | — |
| WeirdML | 39.5% | — |
| LiveBench Coding | — | 39.9% |
Agentic & Tool Use DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 37.9 (#31), Gemma 3 27B: 24.9 (#105)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 56.7% | 29.5% |
| Terminal-Bench | 39.6% | — |
| TheAgentCompany | 42.9% | — |
| Vending-Bench 2 | 1,034 | — |
Reasoning DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 22.0 (#185), Gemma 3 27B: 16.3 (#297)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| Kagi LLM Benchmark | 52.2% | 40.4% |
| CritPt | 2.9% | 0% |
| Chess Puzzles | 14% | 0% |
| LMArena Hard Prompts | 1433 | 1340 |
| DTBench | 87.7% | 52.5% |
| LMCA | 29.1% | 12.3% |
| Epoch Capabilities Index | 146.28 | 130.04 |
| ARC-AGI-2 | 4% | — |
| NYT Connections (extended) | 36.7% | — |
| ARC-AGI-1 | 57% | — |
| Thematic Generalization | 65% | — |
| LiveBench Reasoning | — | 43.8% |
| LiveBench Data Analysis | — | 51.5% |
| LiveBench | — | 50% |
Math DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 40.8 (#75), Gemma 3 27B: 24.4 (#261)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 87.8% | 22.5% |
| LMArena Math | 1435 | 1312 |
| MathArena Final-Answer Competitions | 57.7% | — |
| ProofBench | 8% | — |
| LiveBench Math | — | 55.4% |
| MATH Level 5 | — | 74% |
| FrontierMath (Feb 2025 set) | 22.1% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.6 (#61), Gemma 3 27B: 25.4 (#256)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| GPQA Diamond | 83.4% | 47.7% |
| Vectara Hallucination Rate | 5.3% | 7.4% |
| LMArena Expert | 1436 | 1303 |
| Confabulations | — | 40.3% |
Multimodal Not comparable
DeepSeek-V3.2-Exp: —, Gemma 3 27B: 32.0 (#94)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| LMArena Vision | — | 1165 |
| GeoBench | — | 52% |
Multilingual DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 52.4 (#85), Gemma 3 27B: 47.0 (#150)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| LMArena Non-English | 1409 | 1334 |
| LMArena Chinese | 1461 | 1347 |
| LMArena French | 1435 | 1368 |
| LMArena German | 1440 | 1362 |
| LMArena Japanese | 1377 | 1287 |
| LMArena Korean | 1372 | 1308 |
| LMArena Russian | 1423 | 1349 |
| LMArena Spanish | 1441 | 1348 |
Instruction Following DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 74.7 (#89), Gemma 3 27B: 70.8 (#155)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| LMArena Instruction Following | 1413 | 1321 |
| LiveBench Instruction Following | — | 74.9% |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#15), Gemma 3 27B: 27.6 (#288)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| Fiction.LiveBench | 83.3% | 33.3% |
| LMArena Longer Query | 1428 | 1333 |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 62.5 (#73), Gemma 3 27B: 52.7 (#161)
| Benchmark | DeepSeek-V3.2-Exp | Gemma 3 27B |
|---|---|---|
| LMArena Text | 1425 | 1358 |
| LMArena Creative Writing | 1403 | 1346 |
| EQ-Bench Creative Writing | 1515 | 1266 |
| LMArena Multi-Turn | 1427 | 1344 |
| Short-Story Creative Writing | — | 79.9% |
| LiveBench Language | — | 34.6% |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than Gemma 3 27B?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 30.5 on the Noometry Index. Gemma 3 27B costs 2.9× less per token, which makes it the better buy when DeepSeek-V3.2-Exp's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.2-Exp or Gemma 3 27B?
Gemma 3 27B is cheaper. It lists at $0.08 per million input tokens and $0.16 per million output tokens; DeepSeek-V3.2-Exp lists at $0.26 and $0.38.
Is DeepSeek-V3.2-Exp or Gemma 3 27B better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.4 versus 22.1 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3.2-Exp does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-V3.2-Exp and Gemma 3 27B share?
31 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 48 scored results on Noometry and Gemma 3 27B has 43.