Model comparison
Amazon Nova Pro vs Llama 3.1-70B
Amazon Nova Pro is the stronger model overall, scoring 31.0 to 29.6 on the Noometry Index. Llama 3.1-70B costs 3.5× less per token, which makes it the better buy when Amazon Nova Pro's lead doesn't matter for your workload.
Last verified . 25 shared benchmarks.
MetaSummary
- They share 25 benchmarks with published results for both. Amazon Nova Pro scores higher in 6 categories and Llama 3.1-70B in 3 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where Amazon Nova Pro leads 28.5 to 13.5.
- The biggest single-benchmark swing is TheAgentCompany: 1.7% for Amazon Nova Pro and 6.9% for Llama 3.1-70B.
- Llama 3.1-70B is cheaper at $0.40 / $0.40 per million input/output tokens, against $0.80 / $3.20 for Amazon Nova Pro.
- Amazon Nova Pro accepts more context: 300K tokens versus 128K.
- Llama 3.1-70B has downloadable open weights; the other is API-only.
Side by side
| Amazon Nova Pro | Llama 3.1-70B | |
|---|---|---|
| Provider | Amazon | Meta |
| Noometry Index | 31.0 | 29.6 |
| Released | 2024-12-03 | 2024-07-23 |
| Weights | Proprietary | Open |
| Context window | 300K | 128K |
| Max output | 10K | 4K |
| Input $ / M tokens | $0.80 | $0.40 |
| Output $ / M tokens | $3.20 | $0.40 |
| Results tracked | 38 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Amazon Nova Pro leads
Amazon Nova Pro: 35.1 (#229), Llama 3.1-70B: 30.3 (#296)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| LMArena Coding | 1270 | 1260 |
| WeirdML | — | 9% |
| BigCodeBench Instruct | — | 46.1% |
| LiveBench Coding | 38.1% | — |
| BigCodeBench Complete | — | 54.8% |
Agentic & Tool Use Llama 3.1-70B leads
Amazon Nova Pro: 16.7 (#147), Llama 3.1-70B: 25.1 (#112)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| TheAgentCompany | 1.7% | 6.9% |
| Berkeley Function Calling Leaderboard | 25% | — |
| BALROG | — | 27.9% |
Reasoning Llama 3.1-70B leads
Amazon Nova Pro: 20.0 (#243), Llama 3.1-70B: 21.6 (#220)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1246 | 1241 |
| Epoch Capabilities Index | 123.8 | 125.92 |
| LiveBench Reasoning | 32.6% | — |
| DTBench | — | 60% |
| LiveBench Data Analysis | 48.3% | — |
| LMCA | — | 14.8% |
| LiveBench | 43.5% | — |
Math Amazon Nova Pro leads
Amazon Nova Pro: 28.5 (#243), Llama 3.1-70B: 13.5 (#304)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| Omni-MATH | 24.2% | 21% |
| LMArena Math | 1252 | 1252 |
| OTIS Mock AIME 2024-2025 | — | 3.6% |
| LiveBench Math | 38% | — |
| MATH Level 5 | — | 36.7% |
Knowledge Amazon Nova Pro leads
Amazon Nova Pro: 27.4 (#250), Llama 3.1-70B: 24.2 (#269)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| MMLU-Pro | 67.3% | 65.3% |
| GPQA (HELM) | 44.6% | 42.6% |
| LMArena Expert | 1211 | 1209 |
| MMLU | 82% | 80.1% |
| GPQA Diamond | — | 44.2% |
| Humanity's Last Exam | 4.4% | — |
| Confabulations | 30.1% | — |
| Vectara Hallucination Rate | 5.1% | — |
Multimodal Not comparable
Amazon Nova Pro: 25.0 (#126), Llama 3.1-70B: —
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| LMArena Vision | 980 | — |
Multilingual Too close to call
Amazon Nova Pro: 39.7 (#223), Llama 3.1-70B: 38.8 (#225)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1234 | 1219 |
| LMArena Chinese | 1244 | 1215 |
| LMArena French | 1271 | 1261 |
| LMArena German | 1243 | 1222 |
| LMArena Japanese | 1200 | 1132 |
| LMArena Korean | 1203 | 1140 |
| LMArena Russian | 1240 | 1234 |
| LMArena Spanish | 1182 | 1253 |
Instruction Following Too close to call
Amazon Nova Pro: 64.9 (#226), Llama 3.1-70B: 65.3 (#223)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| IFEval | 81.5% | 82.1% |
| LMArena Instruction Following | 1235 | 1231 |
| LiveBench Instruction Following | 67.1% | — |
Long Context Too close to call
Amazon Nova Pro: 38.1 (#205), Llama 3.1-70B: 37.6 (#214)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1255 | 1241 |
Writing & Preference Amazon Nova Pro leads
Amazon Nova Pro: 43.9 (#226), Llama 3.1-70B: 35.4 (#267)
| Benchmark | Amazon Nova Pro | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1259 | 1261 |
| LMArena Creative Writing | 1212 | 1232 |
| WildBench | 77.7% | 75.8% |
| LMArena Multi-Turn | 1246 | 1256 |
| Short-Story Creative Writing | 60.5% | — |
| EQ-Bench Creative Writing | — | 784 |
| LiveBench Language | 37% | — |
Frequently asked questions
Is Amazon Nova Pro better than Llama 3.1-70B?
Amazon Nova Pro is the stronger model overall, scoring 31.0 to 29.6 on the Noometry Index. Llama 3.1-70B costs 3.5× less per token, which makes it the better buy when Amazon Nova Pro's lead doesn't matter for your workload.
Which is cheaper, Amazon Nova Pro or Llama 3.1-70B?
Llama 3.1-70B is cheaper. It lists at $0.40 per million input tokens and $0.40 per million output tokens; Amazon Nova Pro lists at $0.80 and $3.20.
Is Amazon Nova Pro or Llama 3.1-70B better for coding?
Amazon Nova Pro scores higher on coding benchmarks: 35.1 versus 30.3 in the Noometry coding category.
Which has the bigger context window?
Amazon Nova Pro does, with 300K tokens against 128K.
How many benchmarks do Amazon Nova Pro and Llama 3.1-70B share?
25 benchmarks have published results for both models. Amazon Nova Pro has 38 scored results on Noometry and Llama 3.1-70B has 35.