Model comparison
DeepSeek-V3.2-Speciale vs Llama 3-70B
DeepSeek-V3.2-Speciale is the stronger model overall, scoring 39.1 to 28.5 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 17.4.
Side by side
| DeepSeek-V3.2-Speciale | Llama 3-70B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 39.1 | 28.5 |
| Released | 2025-12-01 | 2024-04-18 |
| Weights | Open | Open |
| Context window | 128K | — |
| Max output | 128K | — |
| Input $ / M tokens | $0.58 | — |
| Output $ / M tokens | $1.68 | — |
| Results tracked | 3 | 31 |
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 3-70B: 35.3 (#215)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| WeirdML | 46.7% | — |
| BigCodeBench Instruct | — | 43.6% |
| LMArena Coding | — | 1206 |
| BigCodeBench Complete | — | 54.5% |
| HumanEval+ | — | 72% |
| MBPP+ | — | 69% |
Agentic & Tool Use Not comparable
DeepSeek-V3.2-Speciale: —, Llama 3-70B: 20.9 (#132)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| Cybench | — | 5% |
Reasoning DeepSeek-V3.2-Speciale leads
DeepSeek-V3.2-Speciale: 32.1 (#65), Llama 3-70B: 17.4 (#284)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| SimpleBench | 52.6% | — |
| Kagi LLM Benchmark | — | 35.1% |
| LMArena Hard Prompts | — | 1195 |
| DTBench | — | 54.2% |
| Epoch Capabilities Index | — | 122.91 |
| ForecastBench | — | 57.1 |
| WinoGrande | — | 83.5% |
Math Not comparable
DeepSeek-V3.2-Speciale: —, Llama 3-70B: 12.0 (#299)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 4.3% |
| LMArena Math | — | 1218 |
| MATH Level 5 | — | 22.6% |
Knowledge Not comparable
DeepSeek-V3.2-Speciale: —, Llama 3-70B: 20.8 (#272)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| GPQA Diamond | — | 40.6% |
| LMArena Expert | — | 1149 |
| MMLU | — | 79.3% |
Multilingual Not comparable
DeepSeek-V3.2-Speciale: —, Llama 3-70B: 33.8 (#246)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| LMArena Non-English | — | 1142 |
| LMArena Chinese | — | 1114 |
| LMArena French | — | 1231 |
| LMArena German | — | 1169 |
| LMArena Japanese | — | 1017 |
| LMArena Korean | — | 1017 |
| LMArena Russian | — | 1159 |
| LMArena Spanish | — | 1241 |
Instruction Following Not comparable
DeepSeek-V3.2-Speciale: —, Llama 3-70B: 62.8 (#233)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| LMArena Instruction Following | — | 1194 |
Long Context Not comparable
DeepSeek-V3.2-Speciale: —, Llama 3-70B: 35.6 (#235)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| LMArena Longer Query | — | 1174 |
Writing & Preference DeepSeek-V3.2-Speciale leads
DeepSeek-V3.2-Speciale: 46.1 (#217), Llama 3-70B: 43.0 (#225)
| Benchmark | DeepSeek-V3.2-Speciale | Llama 3-70B |
|---|---|---|
| LMArena Text | — | 1221 |
| LMArena Creative Writing | — | 1210 |
| EQ-Bench Creative Writing | 1276 | — |
| LMArena Multi-Turn | — | 1223 |
Frequently asked questions
Is DeepSeek-V3.2-Speciale better than Llama 3-70B?
DeepSeek-V3.2-Speciale is the stronger model overall, scoring 39.1 to 28.5 on the Noometry Index.
Is DeepSeek-V3.2-Speciale or Llama 3-70B better for coding?
DeepSeek-V3.2-Speciale scores higher on coding benchmarks: 39.9 versus 35.3 in the Noometry coding category.
How many benchmarks do DeepSeek-V3.2-Speciale and Llama 3-70B share?
0 benchmarks have published results for both models. DeepSeek-V3.2-Speciale has 3 scored results on Noometry and Llama 3-70B has 31.