Model comparison
Amazon Nova Pro vs DeepSeek-R1
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 31.0 on the Noometry Index.
Last verified . 33 shared benchmarks.
DeepSeekSummary
- They share 33 benchmarks with published results for both. Amazon Nova Pro scores higher in 1 category and DeepSeek-R1 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-R1 leads 61.4 to 43.9.
- The biggest single-benchmark swing is LiveBench Reasoning: 32.6% for Amazon Nova Pro and 83.2% for DeepSeek-R1.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $0.80 / $3.20 for Amazon Nova Pro.
- Amazon Nova Pro accepts more context: 300K tokens versus 164K.
Side by side
| Amazon Nova Pro | DeepSeek-R1 | |
|---|---|---|
| Provider | Amazon | DeepSeek |
| Noometry Index | 31.0 | 42.3 |
| Released | 2024-12-03 | 2025-01-20 |
| Weights | Proprietary | Proprietary |
| Context window | 300K | 164K |
| Max output | 10K | 64K |
| Input $ / M tokens | $0.80 | $0.50 |
| Output $ / M tokens | $3.20 | $2.15 |
| Results tracked | 38 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
Amazon Nova Pro: 35.1 (#229), DeepSeek-R1: 46.3 (#68)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| LiveBench Coding | 38.1% | 66.7% |
| LMArena Coding | 1270 | 1427 |
| Aider Polyglot | — | 71.4% |
| SciCode | — | 35.7% |
| WeirdML | — | 41.6% |
| ALE-Bench | — | 804.12 |
| AlgoTune | — | 1.7 |
Agentic & Tool Use DeepSeek-R1 leads
Amazon Nova Pro: 16.7 (#147), DeepSeek-R1: 30.7 (#75)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 25% | — |
| TheAgentCompany | 1.7% | — |
| DeepResearch Bench | — | 35.1% |
| BALROG | — | 34.9% |
| METR Time Horizons | — | 53.8% |
Reasoning Amazon Nova Pro leads
Amazon Nova Pro: 20.0 (#243), DeepSeek-R1: 18.6 (#278)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| LiveBench Reasoning | 32.6% | 83.2% |
| LMArena Hard Prompts | 1246 | 1416 |
| LiveBench Data Analysis | 48.3% | 69.8% |
| Epoch Capabilities Index | 123.8 | 141.29 |
| LiveBench | 43.5% | 71.6% |
| ARC-AGI-2 | — | 1.3% |
| SimpleBench | — | 40.8% |
| Kagi LLM Benchmark | — | 69.4% |
| ARC-AGI-1 | — | 21.2% |
| CritPt | — | 1.1% |
| ForecastBench | — | 60 |
Math DeepSeek-R1 leads
Amazon Nova Pro: 28.5 (#243), DeepSeek-R1: 43.8 (#79)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| Omni-MATH | 24.2% | 42.4% |
| LiveBench Math | 38% | 80.7% |
| LMArena Math | 1252 | 1400 |
| OTIS Mock AIME 2024-2025 | — | 66.4% |
| MATH Level 5 | — | 96.6% |
Knowledge DeepSeek-R1 leads
Amazon Nova Pro: 27.4 (#250), DeepSeek-R1: 44.5 (#87)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| MMLU-Pro | 67.3% | 79.3% |
| Confabulations | 30.1% | 12.7% |
| Vectara Hallucination Rate | 5.1% | 11.3% |
| GPQA (HELM) | 44.6% | 66.6% |
| LMArena Expert | 1211 | 1394 |
| GPQA Diamond | — | 76.3% |
| Humanity's Last Exam | 4.4% | — |
| MMLU | 82% | — |
Multimodal Not comparable
Amazon Nova Pro: 25.0 (#126), DeepSeek-R1: —
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| LMArena Vision | 980 | — |
Multilingual DeepSeek-R1 leads
Amazon Nova Pro: 39.7 (#223), DeepSeek-R1: 52.4 (#85)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| LMArena Non-English | 1234 | 1412 |
| LMArena Chinese | 1244 | 1442 |
| LMArena French | 1271 | 1417 |
| LMArena German | 1243 | 1404 |
| LMArena Japanese | 1200 | 1391 |
| LMArena Korean | 1203 | 1360 |
| LMArena Russian | 1240 | 1423 |
| LMArena Spanish | 1182 | 1411 |
Instruction Following DeepSeek-R1 leads
Amazon Nova Pro: 64.9 (#226), DeepSeek-R1: 72.0 (#143)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| LiveBench Instruction Following | 67.1% | 80.5% |
| IFEval | 81.5% | 78.4% |
| LMArena Instruction Following | 1235 | 1382 |
Long Context DeepSeek-R1 leads
Amazon Nova Pro: 38.1 (#205), DeepSeek-R1: 45.4 (#36)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| LMArena Longer Query | 1255 | 1391 |
| Fiction.LiveBench | — | 75% |
Writing & Preference DeepSeek-R1 leads
Amazon Nova Pro: 43.9 (#226), DeepSeek-R1: 61.4 (#88)
| Benchmark | Amazon Nova Pro | DeepSeek-R1 |
|---|---|---|
| LMArena Text | 1259 | 1428 |
| LMArena Creative Writing | 1212 | 1405 |
| Short-Story Creative Writing | 60.5% | 83% |
| WildBench | 77.7% | 82.8% |
| LMArena Multi-Turn | 1246 | 1405 |
| LiveBench Language | 37% | 48.5% |
| EQ-Bench Creative Writing | — | 1500 |
Frequently asked questions
Is Amazon Nova Pro better than DeepSeek-R1?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 31.0 on the Noometry Index.
Which is cheaper, Amazon Nova Pro or DeepSeek-R1?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Amazon Nova Pro lists at $0.80 and $3.20.
Is Amazon Nova Pro or DeepSeek-R1 better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 35.1 in the Noometry coding category.
Which has the bigger context window?
Amazon Nova Pro does, with 300K tokens against 164K.
How many benchmarks do Amazon Nova Pro and DeepSeek-R1 share?
33 benchmarks have published results for both models. Amazon Nova Pro has 38 scored results on Noometry and DeepSeek-R1 has 52.