# DeepSeek-R1 vs Llama 3.1-8B

> DeepSeek-R1 is the stronger model overall, scoring 42.0 to 22.8 on the Noometry Index. Llama 3.1-8B costs 16× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/deepseek-r1-vs-llama-3-1-8b
- Last updated: 2026-10-10
- Shared benchmarks: 31

## Summary

- They share 31 benchmarks with published results for both. DeepSeek-R1 scores higher in 9 categories and Llama 3.1-8B in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-R1 leads 44.3 to 8.0.
- The biggest single-benchmark swing is MATH Level 5: 96.6% for DeepSeek-R1 and 22.9% for Llama 3.1-8B.
- Llama 3.1-8B is cheaper at $0.05 / $0.08 per million input/output tokens, against $0.50 / $2.15 for DeepSeek-R1.
- DeepSeek-R1 accepts more context: 164K tokens versus 128K.
- Llama 3.1-8B has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 42.0 | 22.8 |
| Rank | 110 | 347 |
| Context | 164K | 128K |
| Input $/M | $0.50 | $0.05 |
| Output $/M | $2.15 | $0.08 |
| Weights | Proprietary | Open |

## Coding

- DeepSeek-R1: 45.9 (#63)
- Llama 3.1-8B: 19.9 (#335)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| SciCode | 35.7% | 13.2% |
| WeirdML | 41.6% | 1.7% |
| LMArena Coding | 1427 | 1195 |
| Aider Polyglot | 71.4% | — |
| BigCodeBench Instruct | — | 32.8% |
| LiveBench Coding | 66.7% | — |
| BigCodeBench Complete | — | 40.5% |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
| HumanEval+ | — | 62.8% |
| MBPP+ | — | 55.6% |

## Agentic & Tool Use

- DeepSeek-R1: 30.4 (#68)
- Llama 3.1-8B: 22.3 (#123)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| BALROG | 34.9% | 15.1% |
| Berkeley Function Calling Leaderboard | — | 25.8% |
| DeepResearch Bench | 35.1% | — |
| METR Time Horizons | 53.8% | — |

## Reasoning

- DeepSeek-R1: 18.7 (#253)
- Llama 3.1-8B: 14.5 (#316)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| CritPt | 1.1% | 0% |
| LMArena Hard Prompts | 1415 | 1175 |
| Epoch Capabilities Index | 141.3 | 116.48 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| Kagi LLM Benchmark | 69.4% | — |
| ARC-AGI-1 | 21.2% | — |
| Chess Puzzles | — | 0% |
| LiveBench Reasoning | 83.2% | — |
| DTBench | — | 50.9% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 5.4% |
| ForecastBench | 60 | — |
| LiveBench | 71.6% | — |
| PIQA | — | 81.2% |

## Math

- DeepSeek-R1: 42.0 (#70)
- Llama 3.1-8B: 9.5 (#311)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 1.7% |
| Omni-MATH | 42.4% | 13.7% |
| LMArena Math | 1401 | 1179 |
| MATH Level 5 | 96.6% | 22.9% |
| LiveBench Math | 80.7% | — |
| GSM8K | — | 82.4% |

## Knowledge

- DeepSeek-R1: 44.3 (#81)
- Llama 3.1-8B: 8.0 (#302)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| GPQA Diamond | 76.3% | 27% |
| MMLU-Pro | 79.3% | 40.6% |
| GPQA (HELM) | 66.6% | 24.7% |
| LMArena Expert | 1394 | 1144 |
| Confabulations | 12.7% | — |
| Vectara Hallucination Rate | 11.3% | — |
| BoolQ | — | 82.8% |
| MMLU | — | 56.1% |

## Multilingual

- DeepSeek-R1: 52.6 (#80)
- Llama 3.1-8B: 34.2 (#244)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| LMArena Non-English | 1412 | 1148 |
| LMArena Chinese | 1443 | 1151 |
| LMArena French | 1415 | 1176 |
| LMArena German | 1405 | 1144 |
| LMArena Japanese | 1394 | 1061 |
| LMArena Korean | 1359 | 1053 |
| LMArena Russian | 1423 | 1158 |
| LMArena Spanish | 1410 | 1169 |

## Instruction Following

- DeepSeek-R1: 72.2 (#138)
- Llama 3.1-8B: 59.1 (#253)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| IFEval | 78.4% | 74.3% |
| LMArena Instruction Following | 1382 | 1159 |
| LiveBench Instruction Following | 80.5% | — |

## Long Context

- DeepSeek-R1: 45.4 (#36)
- Llama 3.1-8B: 35.8 (#233)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| LMArena Longer Query | 1391 | 1182 |
| Fiction.LiveBench | 75% | — |

## Writing & Preference

- DeepSeek-R1: 61.5 (#84)
- Llama 3.1-8B: 29.9 (#285)

| Benchmark | DeepSeek-R1 | Llama 3.1-8B |
|---|---|---|
| LMArena Text | 1428 | 1187 |
| LMArena Creative Writing | 1406 | 1154 |
| EQ-Bench Creative Writing | 1500 | 713 |
| WildBench | 82.8% | 68.7% |
| LMArena Multi-Turn | 1404 | 1172 |
| Short-Story Creative Writing | 83% | — |
| LiveBench Language | 48.5% | — |

## FAQ

### Is DeepSeek-R1 better than Llama 3.1-8B?

DeepSeek-R1 is the stronger model overall, scoring 42.0 to 22.8 on the Noometry Index. Llama 3.1-8B costs 16× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.

### Which is cheaper, DeepSeek-R1 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-R1 lists at $0.50 and $2.15.

### Is DeepSeek-R1 or Llama 3.1-8B better for coding?

DeepSeek-R1 scores higher on coding benchmarks: 45.9 versus 19.9 in the Noometry coding category.

### Which has the bigger context window?

DeepSeek-R1 does, with 164K tokens against 128K.

### How many benchmarks do DeepSeek-R1 and Llama 3.1-8B share?

31 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Llama 3.1-8B has 43.