# DeepSeek-R1 vs Llama 4 Maverick

> DeepSeek-R1 is the stronger model overall, scoring 42.0 to 30.6 on the Noometry Index. Llama 4 Maverick costs 3.0× 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-4-maverick
- Last updated: 2026-10-10
- Shared benchmarks: 41

## Summary

- They share 41 benchmarks with published results for both. DeepSeek-R1 scores higher in 9 categories and Llama 4 Maverick in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-R1 leads 61.5 to 39.0.
- The biggest single-benchmark swing is Aider Polyglot: 71.4% for DeepSeek-R1 and 15.6% for Llama 4 Maverick.
- Llama 4 Maverick is cheaper at $0.19 / $0.65 per million input/output tokens, against $0.50 / $2.15 for DeepSeek-R1.
- DeepSeek-R1 accepts more context: 164K tokens versus 128K.
- Llama 4 Maverick has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 42.0 | 30.6 |
| Rank | 110 | 276 |
| Context | 164K | 128K |
| Input $/M | $0.50 | $0.19 |
| Output $/M | $2.15 | $0.65 |
| Weights | Proprietary | Open |

## Coding

- DeepSeek-R1: 45.9 (#63)
- Llama 4 Maverick: 26.3 (#318)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| Aider Polyglot | 71.4% | 15.6% |
| SciCode | 35.7% | 33.1% |
| WeirdML | 41.6% | 24.5% |
| LMArena Coding | 1427 | 1302 |
| ALE-Bench | 804.12 | 172.97 |
| SWE-bench Verified (bash only) | — | 21% |
| BigCodeBench Instruct | — | 49.7% |
| LiveBench Coding | 66.7% | — |
| BigCodeBench Complete | — | 61.4% |
| AlgoTune | 1.7 | — |

## Agentic & Tool Use

- DeepSeek-R1: 30.4 (#68)
- Llama 4 Maverick: 27.9 (#86)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 37.3% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |

## Reasoning

- DeepSeek-R1: 18.7 (#253)
- Llama 4 Maverick: 10.1 (#337)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| ARC-AGI-2 | 1.3% | 0% |
| SimpleBench | 40.8% | 27.7% |
| Kagi LLM Benchmark | 69.4% | 55.9% |
| ARC-AGI-1 | 21.2% | 4.4% |
| CritPt | 1.1% | 0% |
| LMArena Hard Prompts | 1415 | 1281 |
| Epoch Capabilities Index | 141.3 | 132.2 |
| ForecastBench | 60 | 57.5 |
| NYT Connections (extended) | — | 8% |
| EnigmaEval | — | 0.6% |
| LiveBench Reasoning | 83.2% | — |
| DTBench | — | 61.9% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 15.9% |
| LiveBench | 71.6% | — |

## Math

- DeepSeek-R1: 42.0 (#70)
- Llama 4 Maverick: 24.6 (#256)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 20.6% |
| Omni-MATH | 42.4% | 42.2% |
| LMArena Math | 1401 | 1299 |
| MATH Level 5 | 96.6% | 73% |
| LiveBench Math | 80.7% | — |
| FrontierMath (Feb 2025 set) | — | 0.7% |

## Knowledge

- DeepSeek-R1: 44.3 (#81)
- Llama 4 Maverick: 33.3 (#196)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| GPQA Diamond | 76.3% | 67% |
| MMLU-Pro | 79.3% | 81% |
| Confabulations | 12.7% | 22.6% |
| Vectara Hallucination Rate | 11.3% | 8.2% |
| GPQA (HELM) | 66.6% | 65% |
| LMArena Expert | 1394 | 1259 |
| Humanity's Last Exam | — | 5.7% |

## Multimodal

- DeepSeek-R1: —
- Llama 4 Maverick: 30.9 (#100)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| LMArena Vision | — | 1141 |
| GeoBench | — | 52% |
| SpatialViz-Bench | — | 31.8% |

## Multilingual

- DeepSeek-R1: 52.6 (#80)
- Llama 4 Maverick: 42.4 (#190)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| LMArena Non-English | 1412 | 1269 |
| LMArena Chinese | 1443 | 1277 |
| LMArena French | 1415 | 1258 |
| LMArena German | 1405 | 1291 |
| LMArena Japanese | 1394 | 1207 |
| LMArena Korean | 1359 | 1203 |
| LMArena Russian | 1423 | 1286 |
| LMArena Spanish | 1410 | 1293 |

## Instruction Following

- DeepSeek-R1: 72.2 (#138)
- Llama 4 Maverick: 71.9 (#140)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| IFEval | 78.4% | 90.8% |
| LMArena Instruction Following | 1382 | 1267 |
| LiveBench Instruction Following | 80.5% | — |

## Long Context

- DeepSeek-R1: 45.4 (#36)
- Llama 4 Maverick: 31.4 (#274)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| Fiction.LiveBench | 75% | 46.2% |
| LMArena Longer Query | 1391 | 1280 |

## Writing & Preference

- DeepSeek-R1: 61.5 (#84)
- Llama 4 Maverick: 39.0 (#247)

| Benchmark | DeepSeek-R1 | Llama 4 Maverick |
|---|---|---|
| LMArena Text | 1428 | 1287 |
| LMArena Creative Writing | 1406 | 1266 |
| Short-Story Creative Writing | 83% | 62% |
| EQ-Bench Creative Writing | 1500 | 860 |
| WildBench | 82.8% | 80% |
| LMArena Multi-Turn | 1404 | 1289 |
| LiveBench Language | 48.5% | — |

## FAQ

### Is DeepSeek-R1 better than Llama 4 Maverick?

DeepSeek-R1 is the stronger model overall, scoring 42.0 to 30.6 on the Noometry Index. Llama 4 Maverick costs 3.0× 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 4 Maverick?

Llama 4 Maverick is cheaper. It lists at $0.19 per million input tokens and $0.65 per million output tokens; DeepSeek-R1 lists at $0.50 and $2.15.

### Is DeepSeek-R1 or Llama 4 Maverick better for coding?

DeepSeek-R1 scores higher on coding benchmarks: 45.9 versus 26.3 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 4 Maverick share?

41 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Llama 4 Maverick has 54.