Model comparison
DeepSeek-V2 (MoE-236B, May 2024) vs Llama 3-8B
Llama 3-8B has enough public results to be ranked (#344); DeepSeek-V2 (MoE-236B, May 2024) does not yet, so treat this comparison as directional.
Last verified . 7 shared benchmarks.
DeepSeek
MetaSummary
- They share 7 benchmarks with published results for both. DeepSeek-V2 (MoE-236B, May 2024) scores higher in 1 category and Llama 3-8B in 0 categories; one gap is clear of the uncertainty.
- The widest gap is in coding, where DeepSeek-V2 (MoE-236B, May 2024) leads 40.4 to 31.0.
- The biggest single-benchmark swing is BigCodeBench Complete: 59.4% for DeepSeek-V2 (MoE-236B, May 2024) and 36.9% for Llama 3-8B.
Side by side
| DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 40.3 | 25.5 |
| Released | 2024-05-07 | 2024-04-18 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 10 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V2 (MoE-236B, May 2024) leads
DeepSeek-V2 (MoE-236B, May 2024): 40.4 (#139), Llama 3-8B: 31.0 (#289)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| BigCodeBench Instruct | 48.9% | 31.9% |
| BigCodeBench Complete | 59.4% | 36.9% |
| LMArena Coding | — | 1152 |
| HumanEval+ | — | 56.7% |
| MBPP+ | — | 54.8% |
Reasoning Not comparable
DeepSeek-V2 (MoE-236B, May 2024): —, Llama 3-8B: 14.3 (#326)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| Epoch Capabilities Index | 124.77 | 116.45 |
| WinoGrande | 86.3% | 75.7% |
| Chess Puzzles | — | 0% |
| LMArena Hard Prompts | — | 1133 |
| DTBench | — | 43.9% |
| Adversarial NLI | — | 57.3% |
| BIG-Bench Hard | 78.8% | — |
| ForecastBench | — | 58.6 |
| HellaSwag | 87.1% | — |
| PIQA | 83.9% | — |
Math Not comparable
DeepSeek-V2 (MoE-236B, May 2024): —, Llama 3-8B: 8.8 (#323)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 1.9% |
| LMArena Math | — | 1151 |
| MATH Level 5 | — | 6.1% |
Knowledge Not comparable
DeepSeek-V2 (MoE-236B, May 2024): —, Llama 3-8B: 7.8 (#308)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| ARC (AI2) Challenge | 92.2% | 82.8% |
| MMLU | 78.4% | 68.8% |
| TriviaQA | 80% | 67.7% |
| GPQA Diamond | — | 26.1% |
| LMArena Expert | — | 1113 |
| OpenBookQA | — | 82.6% |
Multilingual Not comparable
DeepSeek-V2 (MoE-236B, May 2024): —, Llama 3-8B: 30.8 (#261)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| LMArena Non-English | — | 1098 |
| LMArena Chinese | — | 1076 |
| LMArena French | — | 1159 |
| LMArena German | — | 1104 |
| LMArena Japanese | — | 967 |
| LMArena Korean | — | 1004 |
| LMArena Russian | — | 1109 |
| LMArena Spanish | — | 1173 |
Instruction Following Not comparable
DeepSeek-V2 (MoE-236B, May 2024): —, Llama 3-8B: 58.4 (#260)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | — | 1127 |
Long Context Not comparable
DeepSeek-V2 (MoE-236B, May 2024): —, Llama 3-8B: 34.2 (#251)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | — | 1128 |
Writing & Preference Not comparable
DeepSeek-V2 (MoE-236B, May 2024): —, Llama 3-8B: 37.5 (#256)
| Benchmark | DeepSeek-V2 (MoE-236B, May 2024) | Llama 3-8B |
|---|---|---|
| LMArena Text | — | 1166 |
| LMArena Creative Writing | — | 1150 |
| LMArena Multi-Turn | — | 1152 |
Frequently asked questions
Is DeepSeek-V2 (MoE-236B, May 2024) better than Llama 3-8B?
Llama 3-8B has enough public results to be ranked (#344); DeepSeek-V2 (MoE-236B, May 2024) does not yet, so treat this comparison as directional.
Is DeepSeek-V2 (MoE-236B, May 2024) or Llama 3-8B better for coding?
DeepSeek-V2 (MoE-236B, May 2024) scores higher on coding benchmarks: 40.4 versus 31.0 in the Noometry coding category.
How many benchmarks do DeepSeek-V2 (MoE-236B, May 2024) and Llama 3-8B share?
7 benchmarks have published results for both models. DeepSeek-V2 (MoE-236B, May 2024) has 10 scored results on Noometry and Llama 3-8B has 34.