Model comparison
Gemma 3 27B vs Llama 3.1-70B
Gemma 3 27B is the stronger model overall, scoring 30.5 to 29.4 on the Noometry Index.
Last verified . 24 shared benchmarks.
Google
MetaSummary
- They share 24 benchmarks with published results for both. Gemma 3 27B scores higher in 5 categories and Llama 3.1-70B in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Gemma 3 27B leads 52.7 to 35.6.
- The biggest single-benchmark swing is MATH Level 5: 74% for Gemma 3 27B and 36.7% for Llama 3.1-70B.
- Gemma 3 27B is cheaper at $0.08 / $0.16 per million input/output tokens, against $0.40 / $0.40 for Llama 3.1-70B.
- Gemma 3 27B accepts more context: 131K tokens versus 128K.
Side by side
| Gemma 3 27B | Llama 3.1-70B | |
|---|---|---|
| Provider | Meta | |
| Noometry Index | 30.5 | 29.4 |
| Released | 2025-03-11 | 2024-07-23 |
| Weights | Open | Open |
| Context window | 131K | 128K |
| Max output | 8K | 4K |
| Input $ / M tokens | $0.08 | $0.40 |
| Output $ / M tokens | $0.16 | $0.40 |
| Results tracked | 43 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 3.1-70B leads
Gemma 3 27B: 22.1 (#329), Llama 3.1-70B: 29.8 (#292)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| LMArena Coding | 1322 | 1260 |
| Aider Polyglot | 4.9% | — |
| SciCode | 21.2% | — |
| WeirdML | — | 9% |
| BigCodeBench Instruct | — | 46.1% |
| LiveBench Coding | 39.9% | — |
| BigCodeBench Complete | — | 54.8% |
Agentic & Tool Use Too close to call
Gemma 3 27B: 24.9 (#105), Llama 3.1-70B: 24.9 (#104)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 29.5% | — |
| TheAgentCompany | — | 6.9% |
| BALROG | — | 27.9% |
Reasoning Llama 3.1-70B leads
Gemma 3 27B: 16.3 (#297), Llama 3.1-70B: 20.8 (#215)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1340 | 1241 |
| DTBench | 52.5% | 60% |
| LMCA | 12.3% | 14.8% |
| Epoch Capabilities Index | 130.04 | 125.91 |
| Kagi LLM Benchmark | 40.4% | — |
| CritPt | 0% | — |
| Chess Puzzles | 0% | — |
| LiveBench Reasoning | 43.8% | — |
| LiveBench Data Analysis | 51.5% | — |
| LiveBench | 50% | — |
Math Gemma 3 27B leads
Gemma 3 27B: 24.4 (#261), Llama 3.1-70B: 12.7 (#298)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 22.5% | 3.6% |
| LMArena Math | 1312 | 1252 |
| MATH Level 5 | 74% | 36.7% |
| Omni-MATH | — | 21% |
| LiveBench Math | 55.4% | — |
Knowledge Gemma 3 27B leads
Gemma 3 27B: 25.4 (#256), Llama 3.1-70B: 24.1 (#264)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| GPQA Diamond | 47.7% | 44.2% |
| LMArena Expert | 1303 | 1209 |
| MMLU-Pro | — | 65.3% |
| Confabulations | 40.3% | — |
| Vectara Hallucination Rate | 7.4% | — |
| GPQA (HELM) | — | 42.6% |
| MMLU | — | 80.1% |
Multimodal Not comparable
Gemma 3 27B: 32.0 (#94), Llama 3.1-70B: —
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| LMArena Vision | 1165 | — |
| GeoBench | 52% | — |
Multilingual Gemma 3 27B leads
Gemma 3 27B: 47.0 (#150), Llama 3.1-70B: 38.9 (#220)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1334 | 1219 |
| LMArena Chinese | 1347 | 1215 |
| LMArena French | 1368 | 1260 |
| LMArena German | 1362 | 1222 |
| LMArena Japanese | 1287 | 1132 |
| LMArena Korean | 1308 | 1140 |
| LMArena Russian | 1349 | 1234 |
| LMArena Spanish | 1348 | 1253 |
Instruction Following Gemma 3 27B leads
Gemma 3 27B: 70.8 (#155), Llama 3.1-70B: 65.5 (#218)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| LMArena Instruction Following | 1321 | 1231 |
| LiveBench Instruction Following | 74.9% | — |
| IFEval | — | 82.1% |
Long Context Llama 3.1-70B leads
Gemma 3 27B: 27.6 (#288), Llama 3.1-70B: 37.6 (#208)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1333 | 1241 |
| Fiction.LiveBench | 33.3% | — |
Writing & Preference Gemma 3 27B leads
Gemma 3 27B: 52.7 (#161), Llama 3.1-70B: 35.6 (#262)
| Benchmark | Gemma 3 27B | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1358 | 1261 |
| LMArena Creative Writing | 1346 | 1232 |
| EQ-Bench Creative Writing | 1266 | 784 |
| LMArena Multi-Turn | 1344 | 1256 |
| Short-Story Creative Writing | 79.9% | — |
| WildBench | — | 75.8% |
| LiveBench Language | 34.6% | — |
Frequently asked questions
Is Gemma 3 27B better than Llama 3.1-70B?
Gemma 3 27B is the stronger model overall, scoring 30.5 to 29.4 on the Noometry Index.
Which is cheaper, Gemma 3 27B or Llama 3.1-70B?
Gemma 3 27B is cheaper. It lists at $0.08 per million input tokens and $0.16 per million output tokens; Llama 3.1-70B lists at $0.40 and $0.40.
Is Gemma 3 27B or Llama 3.1-70B better for coding?
Llama 3.1-70B scores higher on coding benchmarks: 29.8 versus 22.1 in the Noometry coding category.
Which has the bigger context window?
Gemma 3 27B does, with 131K tokens against 128K.
How many benchmarks do Gemma 3 27B and Llama 3.1-70B share?
24 benchmarks have published results for both models. Gemma 3 27B has 43 scored results on Noometry and Llama 3.1-70B has 35.