Model comparison
GPT-5 vs Llama 4 Maverick
GPT-5 is the stronger model overall, scoring 50.9 to 30.9 on the Noometry Index. Llama 4 Maverick costs 11× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 49 shared benchmarks.
MetaSummary
- They share 49 benchmarks with published results for both. GPT-5 scores higher in 10 categories and Llama 4 Maverick in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 31.4.
- The biggest single-benchmark swing is Aider Polyglot: 88% for GPT-5 and 15.6% for Llama 4 Maverick.
- Llama 4 Maverick is cheaper at $0.19 / $0.65 per million input/output tokens, against $1.25 / $10 for GPT-5.
- GPT-5 accepts more context: 400K tokens versus 128K.
- Llama 4 Maverick has downloadable open weights; the other is API-only.
Side by side
| GPT-5 | Llama 4 Maverick | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 50.9 | 30.9 |
| Released | 2025-08-07 | 2025-04-05 |
| Weights | Proprietary | Open |
| Context window | 400K | 128K |
| Max output | 128K | 4K |
| Input $ / M tokens | $1.25 | $0.19 |
| Output $ / M tokens | $10 | $0.65 |
| Results tracked | 69 | 54 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
GPT-5: 50.3 (#47), Llama 4 Maverick: 26.6 (#324)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| SWE-bench Verified (bash only) | 65% | 21% |
| Aider Polyglot | 88% | 15.6% |
| SciCode | 42.9% | 33.1% |
| WeirdML | 60.7% | 24.5% |
| LMArena Coding | 1436 | 1302 |
| ALE-Bench | 1,162 | 172.97 |
| SWE-bench Verified | 73.6% | — |
| LMArena WebDev | 1418 | — |
| GSO | 6.9% | — |
| BigCodeBench Instruct | — | 49.7% |
| BigCodeBench Complete | — | 61.4% |
| AlgoTune | 1.67 | — |
Agentic & Tool Use GPT-5 leads
GPT-5: 33.1 (#56), Llama 4 Maverick: 28.2 (#91)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| Terminal-Bench | 49.6% | — |
| Berkeley Function Calling Leaderboard | — | 37.3% |
| GDPval | 34.8% | — |
| Remote Labor Index | 1.7% | — |
| DeepResearch Bench | 49.6% | — |
| BALROG | 32.8% | — |
| LMArena Search | 1133 | — |
| METR Time Horizons | 69.6% | — |
Reasoning GPT-5 leads
GPT-5: 38.3 (#64), Llama 4 Maverick: 10.1 (#342)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| ARC-AGI-2 | 9.9% | 0% |
| SimpleBench | 56.7% | 27.7% |
| Kagi LLM Benchmark | 72.7% | 55.9% |
| ARC-AGI-1 | 65.7% | 4.4% |
| CritPt | 12.6% | 0% |
| EnigmaEval | 10.5% | 0.6% |
| LMArena Hard Prompts | 1416 | 1281 |
| DTBench | 90.7% | 61.9% |
| LMCA | 40% | 15.9% |
| Epoch Capabilities Index | 150 | 132.2 |
| ForecastBench | 61.4 | 57.5 |
| NYT Connections (extended) | — | 8% |
| Chess Puzzles | 37% | — |
| EBR-Bench | 12.7% | — |
| Mystery Game Puzzles | 23% | — |
Math GPT-5 leads
GPT-5: 55.0 (#44), Llama 4 Maverick: 26.0 (#262)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 91.4% | 20.6% |
| Omni-MATH | 64.7% | 42.2% |
| LMArena Math | 1407 | 1299 |
| MATH Level 5 | 98.1% | 73% |
| FrontierMath (Feb 2025 set) | 32.4% | 0.7% |
| FrontierMath (Tiers 1-3) | 55.4% | — |
| FrontierMath Tier 4 | 22% | — |
| ProofBench | 18% | — |
| FrontierMath Tier 4 (v1) | 12.5% | — |
Knowledge GPT-5 leads
GPT-5: 56.6 (#43), Llama 4 Maverick: 33.4 (#204)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| GPQA Diamond | 86.2% | 67% |
| Humanity's Last Exam | 25.3% | 5.7% |
| MMLU-Pro | 86.3% | 81% |
| Confabulations | 10.3% | 22.6% |
| Vectara Hallucination Rate | 14.7% | 8.2% |
| GPQA (HELM) | 79.2% | 65% |
| LMArena Expert | 1419 | 1259 |
| SimpleQA Verified | 50.1% | — |
Multimodal GPT-5 leads
GPT-5: 46.8 (#13), Llama 4 Maverick: 31.6 (#105)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| LMArena Vision | 1232 | 1142 |
| GeoBench | 81% | 52% |
| VPCT | 66% | — |
| SpatialViz-Bench | — | 31.8% |
Multilingual GPT-5 leads
GPT-5: 51.4 (#110), Llama 4 Maverick: 42.2 (#195)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| LMArena Non-English | 1397 | 1269 |
| LMArena Chinese | 1422 | 1277 |
| LMArena French | 1410 | 1259 |
| LMArena German | 1416 | 1291 |
| LMArena Japanese | 1409 | 1207 |
| LMArena Korean | 1360 | 1203 |
| LMArena Russian | 1406 | 1286 |
| LMArena Spanish | 1399 | 1293 |
Instruction Following GPT-5 leads
GPT-5: 73.8 (#113), Llama 4 Maverick: 71.7 (#146)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| IFEval | 87.5% | 90.8% |
| LMArena Instruction Following | 1388 | 1267 |
Long Context GPT-5 leads
GPT-5: 69.5 (#2), Llama 4 Maverick: 31.4 (#279)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| Fiction.LiveBench | 97.2% | 46.2% |
| LMArena Longer Query | 1399 | 1280 |
Writing & Preference GPT-5 leads
GPT-5: 63.4 (#65), Llama 4 Maverick: 38.8 (#252)
| Benchmark | GPT-5 | Llama 4 Maverick |
|---|---|---|
| LMArena Text | 1406 | 1287 |
| LMArena Creative Writing | 1365 | 1267 |
| Short-Story Creative Writing | 86% | 62% |
| EQ-Bench Creative Writing | 1627 | 860 |
| WildBench | 85.7% | 80% |
| LMArena Multi-Turn | 1426 | 1289 |
Frequently asked questions
Is GPT-5 better than Llama 4 Maverick?
GPT-5 is the stronger model overall, scoring 50.9 to 30.9 on the Noometry Index. Llama 4 Maverick costs 11× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, GPT-5 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-5 lists at $1.25 and $10.
Is GPT-5 or Llama 4 Maverick better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 26.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 128K.
How many benchmarks do GPT-5 and Llama 4 Maverick share?
49 benchmarks have published results for both models. GPT-5 has 69 scored results on Noometry and Llama 4 Maverick has 54.