Model comparison
Gemma 7B vs Llama 3.1-70B
Gemma 7B and Llama 3.1-70B score almost the same on the Noometry Index (29.5 vs 29.4), so choose on price, context window or the category you care about most.
Last verified . 15 shared benchmarks.
Google
MetaSummary
- They share 15 benchmarks with published results for both. Gemma 7B scores higher in 3 categories and Llama 3.1-70B in 5 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Gemma 7B leads 28.7 to 12.7.
Side by side
| Gemma 7B | Llama 3.1-70B | |
|---|---|---|
| Provider | Meta | |
| Noometry Index | 29.5 | 29.4 |
| Released | 2024-02-21 | 2024-07-23 |
| Weights | Open | Open |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.40 |
| Output $ / M tokens | — | $0.40 |
| Results tracked | 27 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Gemma 7B: 29.9 (#290), Llama 3.1-70B: 29.8 (#292)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Coding | 1048 | 1260 |
| WeirdML | — | 9% |
| BigCodeBench Instruct | — | 46.1% |
| BigCodeBench Complete | — | 54.8% |
| HumanEval+ | 28.7% | — |
| MBPP+ | 43.4% | — |
Agentic & Tool Use Not comparable
Gemma 7B: —, Llama 3.1-70B: 24.9 (#104)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| TheAgentCompany | — | 6.9% |
| BALROG | — | 27.9% |
Reasoning Llama 3.1-70B leads
Gemma 7B: 19.0 (#247), Llama 3.1-70B: 20.8 (#215)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1042 | 1241 |
| Epoch Capabilities Index | 111.74 | 125.91 |
| DTBench | — | 60% |
| LMCA | — | 14.8% |
| Adversarial NLI | 48.7% | — |
| BIG-Bench Hard | 55.1% | — |
| HellaSwag | 82.2% | — |
| PIQA | 81.2% | — |
| WinoGrande | 79% | — |
Math Gemma 7B leads
Gemma 7B: 28.7 (#225), Llama 3.1-70B: 12.7 (#298)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Math | 1066 | 1252 |
| OTIS Mock AIME 2024-2025 | — | 3.6% |
| Omni-MATH | — | 21% |
| MATH Level 5 | — | 36.7% |
| GSM8K | 46.4% | — |
Knowledge Gemma 7B leads
Gemma 7B: 26.8 (#248), Llama 3.1-70B: 24.1 (#264)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Expert | 1002 | 1209 |
| MMLU | 66.1% | 80.1% |
| GPQA Diamond | — | 44.2% |
| MMLU-Pro | — | 65.3% |
| GPQA (HELM) | — | 42.6% |
| ARC (AI2) Challenge | 78.3% | — |
| BoolQ | 83.2% | — |
| OpenBookQA | 78.6% | — |
| TriviaQA | 72.3% | — |
Multilingual Llama 3.1-70B leads
Gemma 7B: 25.3 (#282), Llama 3.1-70B: 38.9 (#220)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 999 | 1219 |
| LMArena Chinese | 1035 | 1215 |
| LMArena French | 1024 | 1260 |
| LMArena Russian | 993 | 1234 |
| LMArena German | — | 1222 |
| LMArena Japanese | — | 1132 |
| LMArena Korean | — | 1140 |
| LMArena Spanish | — | 1253 |
Instruction Following Llama 3.1-70B leads
Gemma 7B: 51.7 (#290), Llama 3.1-70B: 65.5 (#218)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Instruction Following | 1017 | 1231 |
| IFEval | — | 82.1% |
Long Context Llama 3.1-70B leads
Gemma 7B: 31.1 (#277), Llama 3.1-70B: 37.6 (#208)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1022 | 1241 |
Writing & Preference Llama 3.1-70B leads
Gemma 7B: 27.2 (#297), Llama 3.1-70B: 35.6 (#262)
| Benchmark | Gemma 7B | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1056 | 1261 |
| LMArena Creative Writing | 1024 | 1232 |
| LMArena Multi-Turn | 963 | 1256 |
| EQ-Bench Creative Writing | — | 784 |
| WildBench | — | 75.8% |
Frequently asked questions
Is Gemma 7B better than Llama 3.1-70B?
Gemma 7B and Llama 3.1-70B score almost the same on the Noometry Index (29.5 vs 29.4), so choose on price, context window or the category you care about most.
Is Gemma 7B or Llama 3.1-70B better for coding?
They score almost the same on coding (29.9 vs 29.8); test both on your own repository before choosing.
How many benchmarks do Gemma 7B and Llama 3.1-70B share?
15 benchmarks have published results for both models. Gemma 7B has 27 scored results on Noometry and Llama 3.1-70B has 35.