Model comparison
DeepSeek-V3.1 vs Llama 3-70B
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 28.8 on the Noometry Index.
Last verified . 21 shared benchmarks.
DeepSeek
MetaSummary
- They share 21 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 8 categories and Llama 3-70B in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V3.1 leads 38.9 to 12.8.
- The biggest single-benchmark swing is DTBench: 82.7% for DeepSeek-V3.1 and 54.2% for Llama 3-70B.
Side by side
| DeepSeek-V3.1 | Llama 3-70B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 42.8 | 28.8 |
| Released | 2025-08-21 | 2024-04-18 |
| Weights | Open | Open |
| Context window | 164K | — |
| Max output | 8K | — |
| Input $ / M tokens | $0.25 | — |
| Output $ / M tokens | $0.95 | — |
| Results tracked | 27 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1 leads
DeepSeek-V3.1: 40.3 (#144), Llama 3-70B: 35.8 (#218)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| LMArena Coding | 1417 | 1206 |
| WeirdML | 38.4% | — |
| BigCodeBench Instruct | — | 43.6% |
| BigCodeBench Complete | — | 54.5% |
| HumanEval+ | — | 72% |
| MBPP+ | — | 69% |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, Llama 3-70B: 21.1 (#139)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| Cybench | — | 5% |
Reasoning DeepSeek-V3.1 leads
DeepSeek-V3.1: 27.9 (#110), Llama 3-70B: 18.0 (#288)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| Kagi LLM Benchmark | 53.2% | 35.1% |
| LMArena Hard Prompts | 1417 | 1195 |
| DTBench | 82.7% | 54.2% |
| Epoch Capabilities Index | 139.92 | 122.93 |
| ForecastBench | 58 | 57.1 |
| SimpleBench | 40% | — |
| LMCA | 24.3% | — |
| WinoGrande | — | 83.5% |
Math DeepSeek-V3.1 leads
DeepSeek-V3.1: 38.9 (#122), Llama 3-70B: 12.8 (#305)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| LMArena Math | 1420 | 1218 |
| OTIS Mock AIME 2024-2025 | — | 4.3% |
| MATH Level 5 | — | 22.6% |
Knowledge DeepSeek-V3.1 leads
DeepSeek-V3.1: 43.7 (#90), Llama 3-70B: 20.8 (#277)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| LMArena Expert | 1405 | 1149 |
| GPQA Diamond | — | 40.6% |
| Vectara Hallucination Rate | 5.5% | — |
| MMLU | — | 79.3% |
Multilingual DeepSeek-V3.1 leads
DeepSeek-V3.1: 51.6 (#106), Llama 3-70B: 33.6 (#251)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| LMArena Non-English | 1400 | 1142 |
| LMArena Chinese | 1469 | 1114 |
| LMArena French | 1447 | 1232 |
| LMArena German | 1411 | 1169 |
| LMArena Japanese | 1378 | 1017 |
| LMArena Korean | 1337 | 1017 |
| LMArena Russian | 1405 | 1159 |
| LMArena Spanish | 1431 | 1241 |
Instruction Following DeepSeek-V3.1 leads
DeepSeek-V3.1: 73.9 (#110), Llama 3-70B: 62.5 (#238)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| LMArena Instruction Following | 1400 | 1194 |
Long Context Too close to call
DeepSeek-V3.1: 36.3 (#232), Llama 3-70B: 35.6 (#240)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| LMArena Longer Query | 1422 | 1174 |
| Fiction.LiveBench | 52.8% | — |
Writing & Preference DeepSeek-V3.1 leads
DeepSeek-V3.1: 60.3 (#98), Llama 3-70B: 42.8 (#231)
| Benchmark | DeepSeek-V3.1 | Llama 3-70B |
|---|---|---|
| LMArena Text | 1420 | 1221 |
| LMArena Creative Writing | 1401 | 1210 |
| LMArena Multi-Turn | 1408 | 1223 |
| EQ-Bench Creative Writing | 1436 | — |
Frequently asked questions
Is DeepSeek-V3.1 better than Llama 3-70B?
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 28.8 on the Noometry Index.
Is DeepSeek-V3.1 or Llama 3-70B better for coding?
DeepSeek-V3.1 scores higher on coding benchmarks: 40.3 versus 35.8 in the Noometry coding category.
How many benchmarks do DeepSeek-V3.1 and Llama 3-70B share?
21 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Llama 3-70B has 31.