Model comparison
DeepSeek-V3 vs Llama 2-13B
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 29.6 on the Noometry Index.
Last verified . 26 shared benchmarks.
DeepSeek
MetaSummary
- They share 26 benchmarks with published results for both. DeepSeek-V3 scores higher in 8 categories and Llama 2-13B in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3 leads 57.4 to 29.8.
- The biggest single-benchmark swing is DTBench: 64.8% for DeepSeek-V3 and 42.2% for Llama 2-13B.
Side by side
| DeepSeek-V3 | Llama 2-13B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 39.5 | 29.6 |
| Released | 2024-12-26 | 2023-07-18 |
| Weights | Open | Open |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 32 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Llama 2-13B: 30.9 (#291)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Coding | 1368 | 1062 |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| WeirdML | 36.1% | — |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Llama 2-13B: —
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| METR Time Horizons | 49.6% | — |
Reasoning DeepSeek-V3 leads
DeepSeek-V3: 20.5 (#236), Llama 2-13B: 12.8 (#337)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Hard Prompts | 1365 | 1051 |
| DTBench | 64.8% | 42.2% |
| BIG-Bench Hard | 87.5% | 58.2% |
| Epoch Capabilities Index | 135.94 | 106.17 |
| HellaSwag | 88.9% | 80.7% |
| PIQA | 84.7% | 80.8% |
| WinoGrande | 85.2% | 72.8% |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | 65.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 15.5% | — |
| ForecastBench | 59.1 | — |
| LAMBADA | — | 76.5% |
| LiveBench | 66.9% | — |
Math Too close to call
DeepSeek-V3: 32.1 (#219), Llama 2-13B: 31.1 (#229)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Math | 1373 | 1065 |
| OTIS Mock AIME 2024-2025 | 37.8% | — |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
| GSM8K | — | 36.9% |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Llama 2-13B: 28.1 (#249)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Expert | 1351 | 1030 |
| ARC (AI2) Challenge | 95.3% | 60.3% |
| MMLU | 87.2% | 55.6% |
| TriviaQA | 82.9% | 79.6% |
| GPQA Diamond | 67.6% | — |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| BoolQ | — | 82.4% |
| OpenBookQA | — | 57% |
Multimodal Not comparable
DeepSeek-V3: —, Llama 2-13B: —
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| ScienceQA | — | 55.8% |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Llama 2-13B: 26.5 (#279)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Non-English | 1358 | 1024 |
| LMArena Chinese | 1391 | 1001 |
| LMArena French | 1385 | 1044 |
| LMArena German | 1374 | 1009 |
| LMArena Japanese | 1333 | 894 |
| LMArena Korean | 1319 | 953 |
| LMArena Russian | 1373 | 1055 |
| LMArena Spanish | 1358 | 1087 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Llama 2-13B: 53.3 (#287)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Instruction Following | 1345 | 1045 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |
Long Context DeepSeek-V3 leads
DeepSeek-V3: 34.0 (#253), Llama 2-13B: 32.3 (#269)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Longer Query | 1352 | 1064 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Llama 2-13B: 29.8 (#289)
| Benchmark | DeepSeek-V3 | Llama 2-13B |
|---|---|---|
| LMArena Text | 1375 | 1084 |
| LMArena Creative Writing | 1364 | 1047 |
| LMArena Multi-Turn | 1389 | 1050 |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Llama 2-13B?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 29.6 on the Noometry Index.
Is DeepSeek-V3 or Llama 2-13B better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 30.9 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and Llama 2-13B share?
26 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Llama 2-13B has 32.