Model comparison
Deepseek Coder v2 vs Llama 3-8B
Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 25.5 on the Noometry Index.
Last verified . 23 shared benchmarks.
DeepSeek
MetaSummary
- They share 23 benchmarks with published results for both. Deepseek Coder v2 scores higher in 8 categories and Llama 3-8B in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Deepseek Coder v2 leads 34.9 to 8.8.
- The biggest single-benchmark swing is BigCodeBench Complete: 59.7% for Deepseek Coder v2 and 36.9% for Llama 3-8B.
Side by side
| Deepseek Coder v2 | Llama 3-8B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 35.9 | 25.5 |
| Released | 2024-06-17 | 2024-04-18 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 24 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Deepseek Coder v2 leads
Deepseek Coder v2: 38.1 (#183), Llama 3-8B: 31.0 (#289)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| BigCodeBench Instruct | 48.2% | 31.9% |
| LMArena Coding | 1251 | 1152 |
| BigCodeBench Complete | 59.7% | 36.9% |
| HumanEval+ | 82.3% | 56.7% |
| MBPP+ | 75.1% | 54.8% |
Reasoning Deepseek Coder v2 leads
Deepseek Coder v2: 23.6 (#176), Llama 3-8B: 14.3 (#326)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| LMArena Hard Prompts | 1207 | 1133 |
| WinoGrande | 83.7% | 75.7% |
| Chess Puzzles | — | 0% |
| DTBench | — | 43.9% |
| Adversarial NLI | — | 57.3% |
| Epoch Capabilities Index | — | 116.45 |
| ForecastBench | — | 58.6 |
Math Deepseek Coder v2 leads
Deepseek Coder v2: 34.9 (#190), Llama 3-8B: 8.8 (#323)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| LMArena Math | 1241 | 1151 |
| OTIS Mock AIME 2024-2025 | — | 1.9% |
| MATH Level 5 | — | 6.1% |
| GSM8K | 94.5% | — |
Knowledge Deepseek Coder v2 leads
Deepseek Coder v2: 32.3 (#212), Llama 3-8B: 7.8 (#308)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| LMArena Expert | 1181 | 1113 |
| ARC (AI2) Challenge | 64.3% | 82.8% |
| GPQA Diamond | — | 26.1% |
| MMLU | — | 68.8% |
| OpenBookQA | — | 82.6% |
| TriviaQA | — | 67.7% |
Multilingual Deepseek Coder v2 leads
Deepseek Coder v2: 36.3 (#240), Llama 3-8B: 30.8 (#261)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| LMArena Non-English | 1182 | 1098 |
| LMArena Chinese | 1201 | 1076 |
| LMArena French | 1185 | 1159 |
| LMArena German | 1164 | 1104 |
| LMArena Japanese | 1126 | 967 |
| LMArena Korean | 1104 | 1004 |
| LMArena Russian | 1188 | 1109 |
| LMArena Spanish | 1153 | 1173 |
Instruction Following Deepseek Coder v2 leads
Deepseek Coder v2: 61.7 (#242), Llama 3-8B: 58.4 (#260)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | 1180 | 1127 |
Long Context Deepseek Coder v2 leads
Deepseek Coder v2: 37.0 (#224), Llama 3-8B: 34.2 (#251)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | 1219 | 1128 |
Writing & Preference Too close to call
Deepseek Coder v2: 38.2 (#253), Llama 3-8B: 37.5 (#256)
| Benchmark | Deepseek Coder v2 | Llama 3-8B |
|---|---|---|
| LMArena Text | 1191 | 1166 |
| LMArena Creative Writing | 1120 | 1150 |
| LMArena Multi-Turn | 1177 | 1152 |
Frequently asked questions
Is Deepseek Coder v2 better than Llama 3-8B?
Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 25.5 on the Noometry Index.
Is Deepseek Coder v2 or Llama 3-8B better for coding?
Deepseek Coder v2 scores higher on coding benchmarks: 38.1 versus 31.0 in the Noometry coding category.
How many benchmarks do Deepseek Coder v2 and Llama 3-8B share?
23 benchmarks have published results for both models. Deepseek Coder v2 has 24 scored results on Noometry and Llama 3-8B has 34.