Model comparison
Amazon Nova Pro vs DeepSeek-V3.2-Exp
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 31.0 on the Noometry Index.
Last verified . 21 shared benchmarks.
DeepSeekSummary
- They share 21 benchmarks with published results for both. Amazon Nova Pro scores higher in 0 categories and DeepSeek-V3.2-Exp in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3.2-Exp leads 51.7 to 27.4.
- The biggest single-benchmark swing is TheAgentCompany: 1.7% for Amazon Nova Pro and 42.9% for DeepSeek-V3.2-Exp.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $0.80 / $3.20 for Amazon Nova Pro.
- Amazon Nova Pro accepts more context: 300K tokens versus 164K.
- DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.
Side by side
| Amazon Nova Pro | DeepSeek-V3.2-Exp | |
|---|---|---|
| Provider | Amazon | DeepSeek |
| Noometry Index | 31.0 | 44.3 |
| Released | 2024-12-03 | 2025-09-29 |
| Weights | Proprietary | Open |
| Context window | 300K | 164K |
| Max output | 10K | 66K |
| Input $ / M tokens | $0.80 | $0.26 |
| Output $ / M tokens | $3.20 | $0.38 |
| Results tracked | 38 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 35.1 (#229), DeepSeek-V3.2-Exp: 46.5 (#65)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Coding | 1270 | 1454 |
| SWE-bench Verified (bash only) | — | 70% |
| Aider Polyglot | — | 74.2% |
| LMArena WebDev | — | 1362 |
| SWE-bench Multilingual | — | 59% |
| SciCode | — | 38.9% |
| WeirdML | — | 39.5% |
| LiveBench Coding | 38.1% | — |
Agentic & Tool Use DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 16.7 (#147), DeepSeek-V3.2-Exp: 32.7 (#59)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| Berkeley Function Calling Leaderboard | 25% | 56.7% |
| TheAgentCompany | 1.7% | 42.9% |
| Terminal-Bench | — | 39.6% |
| APEX-Agents | — | 21.3% |
| Vending-Bench 2 | — | 1,034 |
Reasoning DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 20.0 (#243), DeepSeek-V3.2-Exp: 22.1 (#208)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Hard Prompts | 1246 | 1434 |
| Epoch Capabilities Index | 123.8 | 146.27 |
| ARC-AGI-2 | — | 4% |
| Kagi LLM Benchmark | — | 52.2% |
| NYT Connections (extended) | — | 36.7% |
| ARC-AGI-1 | — | 57% |
| CritPt | — | 2.9% |
| Chess Puzzles | — | 14% |
| Thematic Generalization | — | 65% |
| LiveBench Reasoning | 32.6% | — |
| DTBench | — | 87.7% |
| LiveBench Data Analysis | 48.3% | — |
| LMCA | — | 29.1% |
| LiveBench | 43.5% | — |
Math DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 28.5 (#243), DeepSeek-V3.2-Exp: 41.7 (#87)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Math | 1252 | 1435 |
| MathArena Final-Answer Competitions | — | 57.7% |
| OTIS Mock AIME 2024-2025 | — | 87.8% |
| ProofBench | — | 8% |
| Omni-MATH | 24.2% | — |
| LiveBench Math | 38% | — |
| FrontierMath (Feb 2025 set) | — | 22.1% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 27.4 (#250), DeepSeek-V3.2-Exp: 51.7 (#66)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| Vectara Hallucination Rate | 5.1% | 5.3% |
| LMArena Expert | 1211 | 1436 |
| GPQA Diamond | — | 83.4% |
| Humanity's Last Exam | 4.4% | — |
| MMLU-Pro | 67.3% | — |
| Confabulations | 30.1% | — |
| GPQA (HELM) | 44.6% | — |
| MMLU | 82% | — |
Multimodal Not comparable
Amazon Nova Pro: 25.0 (#126), DeepSeek-V3.2-Exp: —
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Vision | 980 | — |
Multilingual DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 39.7 (#223), DeepSeek-V3.2-Exp: 52.2 (#90)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Non-English | 1234 | 1409 |
| LMArena Chinese | 1244 | 1461 |
| LMArena French | 1271 | 1433 |
| LMArena German | 1243 | 1440 |
| LMArena Japanese | 1200 | 1374 |
| LMArena Korean | 1203 | 1371 |
| LMArena Russian | 1240 | 1424 |
| LMArena Spanish | 1182 | 1440 |
Instruction Following DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 64.9 (#226), DeepSeek-V3.2-Exp: 74.5 (#93)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Instruction Following | 1235 | 1413 |
| LiveBench Instruction Following | 67.1% | — |
| IFEval | 81.5% | — |
Long Context DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 38.1 (#205), DeepSeek-V3.2-Exp: 47.6 (#16)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Longer Query | 1255 | 1428 |
| Fiction.LiveBench | — | 83.3% |
| CL-bench | — | 13.2% |
| CL-bench Life | — | 9.5% |
Writing & Preference DeepSeek-V3.2-Exp leads
Amazon Nova Pro: 43.9 (#226), DeepSeek-V3.2-Exp: 62.4 (#77)
| Benchmark | Amazon Nova Pro | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Text | 1259 | 1425 |
| LMArena Creative Writing | 1212 | 1403 |
| LMArena Multi-Turn | 1246 | 1427 |
| Short-Story Creative Writing | 60.5% | — |
| EQ-Bench Creative Writing | — | 1515 |
| WildBench | 77.7% | — |
| LiveBench Language | 37% | — |
Frequently asked questions
Is Amazon Nova Pro better than DeepSeek-V3.2-Exp?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 31.0 on the Noometry Index.
Which is cheaper, Amazon Nova Pro or DeepSeek-V3.2-Exp?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Amazon Nova Pro lists at $0.80 and $3.20.
Is Amazon Nova Pro or DeepSeek-V3.2-Exp better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.5 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-V3.2-Exp share?
21 benchmarks have published results for both models. Amazon Nova Pro has 38 scored results on Noometry and DeepSeek-V3.2-Exp has 49.