Model comparison
GPT-4o vs Llama-3.3-70B-Instruct
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.3 to 28.5 on the Noometry Index.
Last verified . 41 shared benchmarks.
MetaSummary
- They share 41 benchmarks with published results for both. GPT-4o scores higher in 3 categories and Llama-3.3-70B-Instruct in 6 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-4o leads 39.4 to 26.4.
- The biggest single-benchmark swing is Fiction.LiveBench: 66.7% for GPT-4o and 33.3% for Llama-3.3-70B-Instruct.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.
Side by side
| GPT-4o | Llama-3.3-70B-Instruct | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 28.5 | 30.3 |
| Released | 2024-05-13 | 2024-12-06 |
| Weights | Proprietary | Open |
| Context window | 128K | 128K |
| Max output | 16K | 4K |
| Input $ / M tokens | $2.50 | $0.10 |
| Output $ / M tokens | $10 | $0.32 |
| Results tracked | 72 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama-3.3-70B-Instruct leads
GPT-4o: 25.1 (#323), Llama-3.3-70B-Instruct: 30.6 (#284)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| WeirdML | 25.1% | 14.4% |
| BigCodeBench Instruct | 51.1% | 46.9% |
| LiveBench Coding | 51.4% | 36.6% |
| LMArena Coding | 1297 | 1268 |
| BigCodeBench Complete | 61.1% | 57.5% |
| SWE-bench Verified | 31% | — |
| SWE-bench Verified (bash only) | 21.6% | — |
| Aider Polyglot | 45.3% | — |
| SciCode | — | 26% |
| GSO | 0% | — |
| CadEval | 26% | — |
| HumanEval+ | 87.2% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use Llama-3.3-70B-Instruct leads
GPT-4o: 20.8 (#133), Llama-3.3-70B-Instruct: 25.6 (#100)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| BALROG | 32.3% | 23% |
| Berkeley Function Calling Leaderboard | — | 31.9% |
| GDPval | 9.9% | — |
| TheAgentCompany | 8.6% | — |
| Cybench | 12.5% | — |
| LMArena Search | 1006 | — |
| METR Time Horizons | 40.8% | — |
Reasoning Llama-3.3-70B-Instruct leads
GPT-4o: 9.4 (#338), Llama-3.3-70B-Instruct: 13.7 (#325)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| SimpleBench | 17.8% | 19.9% |
| CritPt | 0% | 0% |
| LiveBench Reasoning | 55.8% | 50.8% |
| LMArena Hard Prompts | 1281 | 1257 |
| DTBench | 64.5% | 59.5% |
| LiveBench Data Analysis | 60.9% | 49.5% |
| LMCA | 16.6% | 17.5% |
| Epoch Capabilities Index | 128.97 | 127.32 |
| ForecastBench | 57.7 | 58.6 |
| LiveBench | 55.3% | 50.2% |
| ARC-AGI-2 | 0% | — |
| ARC-AGI-1 | 4.5% | — |
| Chess Puzzles | 13% | — |
| EnigmaEval | 0.8% | — |
Math Llama-3.3-70B-Instruct leads
GPT-4o: 10.2 (#303), Llama-3.3-70B-Instruct: 14.3 (#292)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.4% | 5.1% |
| LiveBench Math | 49.5% | 42.2% |
| LMArena Math | 1285 | 1268 |
| MATH Level 5 | 53.3% | 41.6% |
| FrontierMath (Tiers 1-3) | 0.4% | — |
| Omni-MATH | 29.3% | — |
| FrontierMath (Feb 2025 set) | 0.3% | — |
Knowledge Llama-3.3-70B-Instruct leads
GPT-4o: 28.8 (#235), Llama-3.3-70B-Instruct: 30.5 (#220)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| GPQA Diamond | 49.2% | 47.4% |
| Confabulations | 15.3% | 22.8% |
| Vectara Hallucination Rate | 9.6% | 4.1% |
| LMArena Expert | 1250 | 1225 |
| MMLU | 88.1% | 86.3% |
| Humanity's Last Exam | 2.7% | — |
| SimpleQA Verified | 26% | — |
| MMLU-Pro | 71.3% | — |
| GPQA (HELM) | 52% | — |
Multimodal Not comparable
GPT-4o: 33.9 (#85), Llama-3.3-70B-Instruct: —
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Vision | 1137 | — |
| Video-MME | 71.9% | — |
| GeoBench | 71% | — |
| VPCT | 40% | — |
| ScienceQA | 88.5% | — |
Multilingual GPT-4o leads
GPT-4o: 43.4 (#181), Llama-3.3-70B-Instruct: 40.1 (#215)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Non-English | 1283 | 1236 |
| LMArena Chinese | 1277 | 1217 |
| LMArena French | 1303 | 1280 |
| LMArena German | 1282 | 1251 |
| LMArena Japanese | 1257 | 1149 |
| LMArena Korean | 1234 | 1142 |
| LMArena Russian | 1286 | 1252 |
| LMArena Spanish | 1292 | 1269 |
Instruction Following Llama-3.3-70B-Instruct leads
GPT-4o: 66.8 (#202), Llama-3.3-70B-Instruct: 71.3 (#152)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| LiveBench Instruction Following | 68.6% | 82.7% |
| LMArena Instruction Following | 1278 | 1242 |
| IFEval | 81.7% | — |
Long Context GPT-4o leads
GPT-4o: 39.4 (#174), Llama-3.3-70B-Instruct: 26.4 (#290)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| Fiction.LiveBench | 66.7% | 33.3% |
| LMArena Longer Query | 1289 | 1256 |
Writing & Preference GPT-4o leads
GPT-4o: 52.8 (#160), Llama-3.3-70B-Instruct: 47.8 (#203)
| Benchmark | GPT-4o | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Text | 1300 | 1274 |
| LMArena Creative Writing | 1291 | 1250 |
| LMArena Multi-Turn | 1302 | 1279 |
| LiveBench Language | 47.6% | 39.2% |
| Short-Story Creative Writing | 81.8% | — |
| WildBench | 82.8% | — |
Frequently asked questions
Is GPT-4o better than Llama-3.3-70B-Instruct?
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.3 to 28.5 on the Noometry Index.
Which is cheaper, GPT-4o or Llama-3.3-70B-Instruct?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; GPT-4o lists at $2.50 and $10.
Is GPT-4o or Llama-3.3-70B-Instruct better for coding?
Llama-3.3-70B-Instruct scores higher on coding benchmarks: 30.6 versus 25.1 in the Noometry coding category.
Which has the bigger context window?
Both accept 128K tokens.
How many benchmarks do GPT-4o and Llama-3.3-70B-Instruct share?
41 benchmarks have published results for both models. GPT-4o has 72 scored results on Noometry and Llama-3.3-70B-Instruct has 43.