Model comparison
GPT-4o vs Llama 3.1-70B
Llama 3.1-70B is the stronger model overall, scoring 29.6 to 28.6 on the Noometry Index.
Last verified . 34 shared benchmarks.
MetaSummary
- They share 34 benchmarks with published results for both. GPT-4o scores higher in 5 categories and Llama 3.1-70B in 4 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GPT-4o leads 52.6 to 35.4.
- The biggest single-benchmark swing is MATH Level 5: 53.3% for GPT-4o and 36.7% for Llama 3.1-70B.
- Llama 3.1-70B is cheaper at $0.40 / $0.40 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- Llama 3.1-70B has downloadable open weights; the other is API-only.
Side by side
| GPT-4o | Llama 3.1-70B | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 28.6 | 29.6 |
| Released | 2024-05-13 | 2024-07-23 |
| Weights | Proprietary | Open |
| Context window | 128K | 128K |
| Max output | 16K | 4K |
| Input $ / M tokens | $2.50 | $0.40 |
| Output $ / M tokens | $10 | $0.40 |
| Results tracked | 72 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 3.1-70B leads
GPT-4o: 24.8 (#328), Llama 3.1-70B: 30.3 (#296)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| WeirdML | 25.1% | 9% |
| BigCodeBench Instruct | 51.1% | 46.1% |
| LMArena Coding | 1297 | 1260 |
| BigCodeBench Complete | 61.1% | 54.8% |
| SWE-bench Verified | 31% | — |
| SWE-bench Verified (bash only) | 21.6% | — |
| Aider Polyglot | 45.3% | — |
| GSO | 0% | — |
| LiveBench Coding | 51.4% | — |
| CadEval | 26% | — |
| HumanEval+ | 87.2% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use Llama 3.1-70B leads
GPT-4o: 21.0 (#141), Llama 3.1-70B: 25.1 (#112)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| TheAgentCompany | 8.6% | 6.9% |
| BALROG | 32.3% | 27.9% |
| GDPval | 9.9% | — |
| Cybench | 12.5% | — |
| LMArena Search | 1006 | — |
| METR Time Horizons | 40.8% | — |
Reasoning Llama 3.1-70B leads
GPT-4o: 9.4 (#343), Llama 3.1-70B: 21.6 (#220)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1281 | 1241 |
| DTBench | 64.5% | 60% |
| LMCA | 16.6% | 14.8% |
| Epoch Capabilities Index | 128.97 | 125.92 |
| ARC-AGI-2 | 0% | — |
| SimpleBench | 17.8% | — |
| ARC-AGI-1 | 4.5% | — |
| CritPt | 0% | — |
| Chess Puzzles | 13% | — |
| EnigmaEval | 0.8% | — |
| LiveBench Reasoning | 55.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| ForecastBench | 57.7 | — |
| LiveBench | 55.3% | — |
Math Llama 3.1-70B leads
GPT-4o: 10.6 (#312), Llama 3.1-70B: 13.5 (#304)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.4% | 3.6% |
| Omni-MATH | 29.3% | 21% |
| LMArena Math | 1285 | 1252 |
| MATH Level 5 | 53.3% | 36.7% |
| FrontierMath (Tiers 1-3) | 0.4% | — |
| LiveBench Math | 49.5% | — |
| FrontierMath (Feb 2025 set) | 0.3% | — |
Knowledge GPT-4o leads
GPT-4o: 28.8 (#242), Llama 3.1-70B: 24.2 (#269)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| GPQA Diamond | 49.2% | 44.2% |
| MMLU-Pro | 71.3% | 65.3% |
| GPQA (HELM) | 52% | 42.6% |
| LMArena Expert | 1250 | 1209 |
| MMLU | 88.1% | 80.1% |
| Humanity's Last Exam | 2.7% | — |
| SimpleQA Verified | 26% | — |
| Confabulations | 15.3% | — |
| Vectara Hallucination Rate | 9.6% | — |
Multimodal Not comparable
GPT-4o: 34.5 (#91), Llama 3.1-70B: —
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| LMArena Vision | 1137 | — |
| Video-MME | 71.9% | — |
| GeoBench | 71% | — |
| VPCT | 40% | — |
| ScienceQA | 88.5% | — |
Multilingual GPT-4o leads
GPT-4o: 43.2 (#186), Llama 3.1-70B: 38.8 (#225)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1283 | 1219 |
| LMArena Chinese | 1277 | 1215 |
| LMArena French | 1304 | 1261 |
| LMArena German | 1282 | 1222 |
| LMArena Japanese | 1257 | 1132 |
| LMArena Korean | 1234 | 1140 |
| LMArena Russian | 1286 | 1234 |
| LMArena Spanish | 1292 | 1253 |
Instruction Following GPT-4o leads
GPT-4o: 66.6 (#207), Llama 3.1-70B: 65.3 (#223)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| IFEval | 81.7% | 82.1% |
| LMArena Instruction Following | 1278 | 1231 |
| LiveBench Instruction Following | 68.6% | — |
Long Context GPT-4o leads
GPT-4o: 39.4 (#179), Llama 3.1-70B: 37.6 (#214)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1289 | 1241 |
| Fiction.LiveBench | 66.7% | — |
Writing & Preference GPT-4o leads
GPT-4o: 52.6 (#166), Llama 3.1-70B: 35.4 (#267)
| Benchmark | GPT-4o | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1300 | 1261 |
| LMArena Creative Writing | 1292 | 1232 |
| WildBench | 82.8% | 75.8% |
| LMArena Multi-Turn | 1302 | 1256 |
| Short-Story Creative Writing | 81.8% | — |
| EQ-Bench Creative Writing | — | 784 |
| LiveBench Language | 47.6% | — |
Frequently asked questions
Is GPT-4o better than Llama 3.1-70B?
Llama 3.1-70B is the stronger model overall, scoring 29.6 to 28.6 on the Noometry Index.
Which is cheaper, GPT-4o 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; GPT-4o lists at $2.50 and $10.
Is GPT-4o or Llama 3.1-70B better for coding?
Llama 3.1-70B scores higher on coding benchmarks: 30.3 versus 24.8 in the Noometry coding category.
Which has the bigger context window?
Both accept 128K tokens.
How many benchmarks do GPT-4o and Llama 3.1-70B share?
34 benchmarks have published results for both models. GPT-4o has 72 scored results on Noometry and Llama 3.1-70B has 35.