Model comparison
DeepSeek-V3.2-Speciale vs Llama 2-70B
DeepSeek-V3.2-Speciale is the stronger model overall, scoring 39.1 to 24.1 on the Noometry Index.
Last verified . 0 shared benchmarks.
DeepSeek
MetaSummary
- The widest gap is in reasoning, where DeepSeek-V3.2-Speciale leads 32.1 to 13.7.
Side by side
| DeepSeek-V3.2-Speciale | Llama 2-70B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 39.1 | 24.1 |
| Released | 2025-12-01 | 2023-07-18 |
| Weights | Open | Open |
| Context window | 128K | — |
| Max output | 128K | — |
| Input $ / M tokens | $0.58 | — |
| Output $ / M tokens | $1.68 | — |
| Results tracked | 3 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Speciale leads
DeepSeek-V3.2-Speciale: 39.9 (#136), Llama 2-70B: 30.8 (#281)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| WeirdML | 46.7% | — |
| LMArena Coding | — | 1079 |
Reasoning DeepSeek-V3.2-Speciale leads
DeepSeek-V3.2-Speciale: 32.1 (#65), Llama 2-70B: 13.7 (#324)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| SimpleBench | 52.6% | — |
| LMArena Hard Prompts | — | 1073 |
| DTBench | — | 41.6% |
| BIG-Bench Hard | — | 64.9% |
| CommonsenseQA 2.0 | — | 50% |
| Epoch Capabilities Index | — | 113.59 |
| ForecastBench | — | 51.4 |
| HellaSwag | — | 85.3% |
| LAMBADA | — | 78.9% |
| PIQA | — | 82.8% |
| WinoGrande | — | 80.2% |
Math Not comparable
DeepSeek-V3.2-Speciale: —, Llama 2-70B: 7.5 (#320)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 0% |
| LMArena Math | — | 1091 |
| MATH Level 5 | — | 3.3% |
| GSM8K | — | 69.6% |
Knowledge Not comparable
DeepSeek-V3.2-Speciale: —, Llama 2-70B: 7.4 (#305)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| GPQA Diamond | — | 26.3% |
| LMArena Expert | — | 1039 |
| ARC (AI2) Challenge | — | 78.3% |
| BoolQ | — | 88.6% |
| MMLU | — | 69.9% |
| OpenBookQA | — | 60.2% |
| TriviaQA | — | 87.6% |
Multilingual Not comparable
DeepSeek-V3.2-Speciale: —, Llama 2-70B: 27.9 (#269)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| LMArena Non-English | — | 1045 |
| LMArena Chinese | — | 994 |
| LMArena French | — | 1089 |
| LMArena German | — | 1041 |
| LMArena Japanese | — | 927 |
| LMArena Korean | — | 964 |
| LMArena Russian | — | 1083 |
| LMArena Spanish | — | 1142 |
Instruction Following Not comparable
DeepSeek-V3.2-Speciale: —, Llama 2-70B: 55.1 (#273)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| LMArena Instruction Following | — | 1071 |
Long Context Not comparable
DeepSeek-V3.2-Speciale: —, Llama 2-70B: 32.2 (#265)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| LMArena Longer Query | — | 1062 |
Writing & Preference DeepSeek-V3.2-Speciale leads
DeepSeek-V3.2-Speciale: 46.1 (#217), Llama 2-70B: 32.5 (#274)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 2-70B |
|---|---|---|
| LMArena Text | — | 1115 |
| LMArena Creative Writing | — | 1075 |
| EQ-Bench Creative Writing | 1276 | — |
| LMArena Multi-Turn | — | 1088 |
Frequently asked questions
Is DeepSeek-V3.2-Speciale better than Llama 2-70B?
DeepSeek-V3.2-Speciale is the stronger model overall, scoring 39.1 to 24.1 on the Noometry Index.
Is DeepSeek-V3.2-Speciale or Llama 2-70B better for coding?
DeepSeek-V3.2-Speciale scores higher on coding benchmarks: 39.9 versus 30.8 in the Noometry coding category.
How many benchmarks do DeepSeek-V3.2-Speciale and Llama 2-70B share?
0 benchmarks have published results for both models. DeepSeek-V3.2-Speciale has 3 scored results on Noometry and Llama 2-70B has 35.