Model comparison
GPT-4o vs Llama 4 Maverick
Llama 4 Maverick is the stronger model overall, scoring 30.9 to 28.6 on the Noometry Index.
Last verified . 47 shared benchmarks.
MetaSummary
- They share 47 benchmarks with published results for both. GPT-4o scores higher in 4 categories and Llama 4 Maverick in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Llama 4 Maverick leads 26.0 to 10.6.
- The biggest single-benchmark swing is Aider Polyglot: 45.3% for GPT-4o and 15.6% for Llama 4 Maverick.
- Llama 4 Maverick is cheaper at $0.19 / $0.65 per million input/output tokens, against $2.50 / $10 for GPT-4o.
- Llama 4 Maverick has downloadable open weights; the other is API-only.
Side by side
| GPT-4o | Llama 4 Maverick | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 28.6 | 30.9 |
| Released | 2024-05-13 | 2025-04-05 |
| Weights | Proprietary | Open |
| Context window | 128K | 128K |
| Max output | 16K | 4K |
| Input $ / M tokens | $2.50 | $0.19 |
| Output $ / M tokens | $10 | $0.65 |
| Results tracked | 72 | 54 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama 4 Maverick leads
GPT-4o: 24.8 (#328), Llama 4 Maverick: 26.6 (#324)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| SWE-bench Verified (bash only) | 21.6% | 21% |
| Aider Polyglot | 45.3% | 15.6% |
| WeirdML | 25.1% | 24.5% |
| BigCodeBench Instruct | 51.1% | 49.7% |
| LMArena Coding | 1297 | 1302 |
| BigCodeBench Complete | 61.1% | 61.4% |
| SWE-bench Verified | 31% | — |
| SciCode | — | 33.1% |
| GSO | 0% | — |
| LiveBench Coding | 51.4% | — |
| CadEval | 26% | — |
| ALE-Bench | — | 172.97 |
| HumanEval+ | 87.2% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use Llama 4 Maverick leads
GPT-4o: 21.0 (#141), Llama 4 Maverick: 28.2 (#91)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 37.3% |
| GDPval | 9.9% | — |
| TheAgentCompany | 8.6% | — |
| Cybench | 12.5% | — |
| BALROG | 32.3% | — |
| LMArena Search | 1006 | — |
| METR Time Horizons | 40.8% | — |
Reasoning Too close to call
GPT-4o: 9.4 (#343), Llama 4 Maverick: 10.1 (#342)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| ARC-AGI-2 | 0% | 0% |
| SimpleBench | 17.8% | 27.7% |
| ARC-AGI-1 | 4.5% | 4.4% |
| CritPt | 0% | 0% |
| EnigmaEval | 0.8% | 0.6% |
| LMArena Hard Prompts | 1281 | 1281 |
| DTBench | 64.5% | 61.9% |
| LMCA | 16.6% | 15.9% |
| Epoch Capabilities Index | 128.97 | 132.2 |
| ForecastBench | 57.7 | 57.5 |
| Kagi LLM Benchmark | — | 55.9% |
| NYT Connections (extended) | — | 8% |
| Chess Puzzles | 13% | — |
| LiveBench Reasoning | 55.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LiveBench | 55.3% | — |
Math Llama 4 Maverick leads
GPT-4o: 10.6 (#312), Llama 4 Maverick: 26.0 (#262)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.4% | 20.6% |
| Omni-MATH | 29.3% | 42.2% |
| LMArena Math | 1285 | 1299 |
| MATH Level 5 | 53.3% | 73% |
| FrontierMath (Feb 2025 set) | 0.3% | 0.7% |
| FrontierMath (Tiers 1-3) | 0.4% | — |
| LiveBench Math | 49.5% | — |
Knowledge Llama 4 Maverick leads
GPT-4o: 28.8 (#242), Llama 4 Maverick: 33.4 (#204)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| GPQA Diamond | 49.2% | 67% |
| Humanity's Last Exam | 2.7% | 5.7% |
| MMLU-Pro | 71.3% | 81% |
| Confabulations | 15.3% | 22.6% |
| Vectara Hallucination Rate | 9.6% | 8.2% |
| GPQA (HELM) | 52% | 65% |
| LMArena Expert | 1250 | 1259 |
| SimpleQA Verified | 26% | — |
| MMLU | 88.1% | — |
Multimodal GPT-4o leads
GPT-4o: 34.5 (#91), Llama 4 Maverick: 31.6 (#105)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| LMArena Vision | 1137 | 1142 |
| GeoBench | 71% | 52% |
| Video-MME | 71.9% | — |
| VPCT | 40% | — |
| ScienceQA | 88.5% | — |
| SpatialViz-Bench | — | 31.8% |
Multilingual Too close to call
GPT-4o: 43.2 (#186), Llama 4 Maverick: 42.2 (#195)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| LMArena Non-English | 1283 | 1269 |
| LMArena Chinese | 1277 | 1277 |
| LMArena French | 1304 | 1259 |
| LMArena German | 1282 | 1291 |
| LMArena Japanese | 1257 | 1207 |
| LMArena Korean | 1234 | 1203 |
| LMArena Russian | 1286 | 1286 |
| LMArena Spanish | 1292 | 1293 |
Instruction Following Llama 4 Maverick leads
GPT-4o: 66.6 (#207), Llama 4 Maverick: 71.7 (#146)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| IFEval | 81.7% | 90.8% |
| LMArena Instruction Following | 1278 | 1267 |
| LiveBench Instruction Following | 68.6% | — |
Long Context GPT-4o leads
GPT-4o: 39.4 (#179), Llama 4 Maverick: 31.4 (#279)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| Fiction.LiveBench | 66.7% | 46.2% |
| LMArena Longer Query | 1289 | 1280 |
Writing & Preference GPT-4o leads
GPT-4o: 52.6 (#166), Llama 4 Maverick: 38.8 (#252)
| Benchmark | GPT-4o | Llama 4 Maverick |
|---|---|---|
| LMArena Text | 1300 | 1287 |
| LMArena Creative Writing | 1292 | 1267 |
| Short-Story Creative Writing | 81.8% | 62% |
| WildBench | 82.8% | 80% |
| LMArena Multi-Turn | 1302 | 1289 |
| EQ-Bench Creative Writing | — | 860 |
| LiveBench Language | 47.6% | — |
Frequently asked questions
Is GPT-4o better than Llama 4 Maverick?
Llama 4 Maverick is the stronger model overall, scoring 30.9 to 28.6 on the Noometry Index.
Which is cheaper, GPT-4o or Llama 4 Maverick?
Llama 4 Maverick is cheaper. It lists at $0.19 per million input tokens and $0.65 per million output tokens; GPT-4o lists at $2.50 and $10.
Is GPT-4o or Llama 4 Maverick better for coding?
Llama 4 Maverick scores higher on coding benchmarks: 26.6 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 4 Maverick share?
47 benchmarks have published results for both models. GPT-4o has 72 scored results on Noometry and Llama 4 Maverick has 54.