Model comparison
DeepSeek-V3 vs Gemma 4 26B A4B IT
Gemma 4 26B A4B IT is the stronger model overall, scoring 43.5 to 39.5 on the Noometry Index.
Last verified . 24 shared benchmarks.
DeepSeek
GoogleSummary
- They share 24 benchmarks with published results for both. DeepSeek-V3 scores higher in 1 category and Gemma 4 26B A4B IT in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 4 26B A4B IT leads 47.6 to 32.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 82.2% for Gemma 4 26B A4B IT.
- Gemma 4 26B A4B IT is cheaper at $0.0675 / $0.23 per million input/output tokens, against $0.24 / $0.90 for DeepSeek-V3.
- Gemma 4 26B A4B IT accepts more context: 262K tokens versus 164K.
Side by side
| DeepSeek-V3 | Gemma 4 26B A4B IT | |
|---|---|---|
| Provider | DeepSeek | |
| Noometry Index | 39.5 | 43.5 |
| Released | 2024-12-26 | 2026-04-02 |
| Weights | Open | Open |
| Context window | 164K | 262K |
| Max output | 164K | 33K |
| Input $ / M tokens | $0.24 | $0.0675 |
| Output $ / M tokens | $0.90 | $0.23 |
| Results tracked | 60 | 28 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Gemma 4 26B A4B IT: 39.0 (#164)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| SciCode | 35.8% | 40% |
| WeirdML | 36.1% | 35.2% |
| LMArena Coding | 1368 | 1447 |
| Aider Polyglot | 55.1% | — |
| LMArena WebDev | — | 1359 |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| ALE-Bench | — | 927.17 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Gemma 4 26B A4B IT: —
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| METR Time Horizons | 49.6% | — |
Reasoning Gemma 4 26B A4B IT leads
DeepSeek-V3: 20.5 (#236), Gemma 4 26B A4B IT: 21.8 (#213)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| CritPt | 0% | 0% |
| LMArena Hard Prompts | 1365 | 1439 |
| DTBench | 64.8% | 74.9% |
| LMCA | 15.5% | 29.7% |
| Epoch Capabilities Index | 135.94 | 141.85 |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| Chess Puzzles | — | 6% |
| LiveBench Reasoning | 65.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| BIG-Bench Hard | 87.5% | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math Gemma 4 26B A4B IT leads
DeepSeek-V3: 32.1 (#219), Gemma 4 26B A4B IT: 47.6 (#67)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 82.2% |
| LMArena Math | 1373 | 1470 |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge Gemma 4 26B A4B IT leads
DeepSeek-V3: 37.5 (#155), Gemma 4 26B A4B IT: 45.7 (#85)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| GPQA Diamond | 67.6% | 73.2% |
| Vectara Hallucination Rate | 6.1% | 5.2% |
| LMArena Expert | 1351 | 1447 |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multimodal Not comparable
DeepSeek-V3: —, Gemma 4 26B A4B IT: 40.6 (#46)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| LMArena Vision | — | 1260 |
Multilingual Gemma 4 26B A4B IT leads
DeepSeek-V3: 48.5 (#143), Gemma 4 26B A4B IT: 53.1 (#71)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| LMArena Non-English | 1358 | 1421 |
| LMArena Chinese | 1391 | 1495 |
| LMArena French | 1385 | 1460 |
| LMArena Russian | 1373 | 1434 |
| LMArena Spanish | 1358 | 1417 |
| LMArena German | 1374 | — |
| LMArena Japanese | 1333 | — |
| LMArena Korean | 1319 | — |
Instruction Following Gemma 4 26B A4B IT leads
DeepSeek-V3: 72.8 (#130), Gemma 4 26B A4B IT: 74.8 (#82)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| LMArena Instruction Following | 1345 | 1420 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |
Long Context Gemma 4 26B A4B IT leads
DeepSeek-V3: 34.0 (#253), Gemma 4 26B A4B IT: 43.6 (#91)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| LMArena Longer Query | 1352 | 1428 |
| Fiction.LiveBench | 50% | — |
Writing & Preference Gemma 4 26B A4B IT leads
DeepSeek-V3: 57.4 (#130), Gemma 4 26B A4B IT: 58.6 (#115)
| Benchmark | DeepSeek-V3 | Gemma 4 26B A4B IT |
|---|---|---|
| LMArena Text | 1375 | 1434 |
| LMArena Creative Writing | 1364 | 1402 |
| EQ-Bench Creative Writing | 1472 | 1305 |
| LMArena Multi-Turn | 1389 | 1441 |
| Short-Story Creative Writing | 77% | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Gemma 4 26B A4B IT?
Gemma 4 26B A4B IT is the stronger model overall, scoring 43.5 to 39.5 on the Noometry Index.
Which is cheaper, DeepSeek-V3 or Gemma 4 26B A4B IT?
Gemma 4 26B A4B IT is cheaper. It lists at $0.0675 per million input tokens and $0.23 per million output tokens; DeepSeek-V3 lists at $0.24 and $0.90.
Is DeepSeek-V3 or Gemma 4 26B A4B IT better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 39.0 in the Noometry coding category.
Which has the bigger context window?
Gemma 4 26B A4B IT does, with 262K tokens against 164K.
How many benchmarks do DeepSeek-V3 and Gemma 4 26B A4B IT share?
24 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Gemma 4 26B A4B IT has 28.