Model comparison
DeepSeek-V3.1 vs Llama 3.1-8B
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 23.0 on the Noometry Index. Llama 3.1-8B costs 7.4× less per token, which makes it the better buy when DeepSeek-V3.1's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
DeepSeek
MetaSummary
- They share 22 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 8 categories and Llama 3.1-8B in 0 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3.1 leads 43.7 to 8.0.
- The biggest single-benchmark swing is WeirdML: 38.4% for DeepSeek-V3.1 and 1.7% for Llama 3.1-8B.
- Llama 3.1-8B is cheaper at $0.05 / $0.08 per million input/output tokens, against $0.25 / $0.95 for DeepSeek-V3.1.
- DeepSeek-V3.1 accepts more context: 164K tokens versus 128K.
Side by side
| DeepSeek-V3.1 | Llama 3.1-8B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 42.8 | 23.0 |
| Released | 2025-08-21 | 2024-07-23 |
| Weights | Open | Open |
| Context window | 164K | 128K |
| Max output | 8K | 4K |
| Input $ / M tokens | $0.25 | $0.05 |
| Output $ / M tokens | $0.95 | $0.08 |
| Results tracked | 27 | 43 |
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 3.1-8B: 20.2 (#340)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| WeirdML | 38.4% | 1.7% |
| LMArena Coding | 1417 | 1195 |
| SciCode | — | 13.2% |
| BigCodeBench Instruct | — | 32.8% |
| BigCodeBench Complete | — | 40.5% |
| HumanEval+ | — | 62.8% |
| MBPP+ | — | 55.6% |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, Llama 3.1-8B: 22.5 (#131)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 25.8% |
| BALROG | — | 15.1% |
Reasoning DeepSeek-V3.1 leads
DeepSeek-V3.1: 27.9 (#110), Llama 3.1-8B: 14.9 (#321)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| LMArena Hard Prompts | 1417 | 1175 |
| DTBench | 82.7% | 50.9% |
| LMCA | 24.3% | 5.4% |
| Epoch Capabilities Index | 139.92 | 116.57 |
| SimpleBench | 40% | — |
| Kagi LLM Benchmark | 53.2% | — |
| CritPt | — | 0% |
| Chess Puzzles | — | 0% |
| ForecastBench | 58 | — |
| PIQA | — | 81.2% |
Math DeepSeek-V3.1 leads
DeepSeek-V3.1: 38.9 (#122), Llama 3.1-8B: 10.2 (#317)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| LMArena Math | 1420 | 1179 |
| OTIS Mock AIME 2024-2025 | — | 1.7% |
| Omni-MATH | — | 13.7% |
| MATH Level 5 | — | 22.9% |
| GSM8K | — | 82.4% |
Knowledge DeepSeek-V3.1 leads
DeepSeek-V3.1: 43.7 (#90), Llama 3.1-8B: 8.0 (#307)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| LMArena Expert | 1405 | 1144 |
| GPQA Diamond | — | 27% |
| MMLU-Pro | — | 40.6% |
| Vectara Hallucination Rate | 5.5% | — |
| GPQA (HELM) | — | 24.7% |
| BoolQ | — | 82.8% |
| MMLU | — | 56.1% |
Multilingual DeepSeek-V3.1 leads
DeepSeek-V3.1: 51.6 (#106), Llama 3.1-8B: 34.0 (#249)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| LMArena Non-English | 1400 | 1148 |
| LMArena Chinese | 1469 | 1151 |
| LMArena French | 1447 | 1177 |
| LMArena German | 1411 | 1144 |
| LMArena Japanese | 1378 | 1061 |
| LMArena Korean | 1337 | 1053 |
| LMArena Russian | 1405 | 1158 |
| LMArena Spanish | 1431 | 1169 |
Instruction Following DeepSeek-V3.1 leads
DeepSeek-V3.1: 73.9 (#110), Llama 3.1-8B: 58.9 (#258)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| LMArena Instruction Following | 1400 | 1159 |
| IFEval | — | 74.3% |
Long Context Too close to call
DeepSeek-V3.1: 36.3 (#232), Llama 3.1-8B: 35.8 (#238)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| LMArena Longer Query | 1422 | 1182 |
| Fiction.LiveBench | 52.8% | — |
Writing & Preference DeepSeek-V3.1 leads
DeepSeek-V3.1: 60.3 (#98), Llama 3.1-8B: 29.7 (#290)
| Benchmark | DeepSeek-V3.1 | Llama 3.1-8B |
|---|---|---|
| LMArena Text | 1420 | 1187 |
| LMArena Creative Writing | 1401 | 1154 |
| EQ-Bench Creative Writing | 1436 | 713 |
| LMArena Multi-Turn | 1408 | 1172 |
| WildBench | — | 68.7% |
Frequently asked questions
Is DeepSeek-V3.1 better than Llama 3.1-8B?
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 23.0 on the Noometry Index. Llama 3.1-8B costs 7.4× less per token, which makes it the better buy when DeepSeek-V3.1's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.1 or Llama 3.1-8B?
Llama 3.1-8B is cheaper. It lists at $0.05 per million input tokens and $0.08 per million output tokens; DeepSeek-V3.1 lists at $0.25 and $0.95.
Is DeepSeek-V3.1 or Llama 3.1-8B better for coding?
DeepSeek-V3.1 scores higher on coding benchmarks: 40.3 versus 20.2 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3.1 does, with 164K tokens against 128K.
How many benchmarks do DeepSeek-V3.1 and Llama 3.1-8B share?
22 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Llama 3.1-8B has 43.