Rankings / Meta
Llama 4 Maverick
released 2025-04-05
BenchAtlas Index
as of 2026-07-14
base configuration
rank #402
12 families · 6 categories · high
Benchmark evidence
19 results
agentic coding
| τ²-Bench | 17.8 % | independent | Artificial Analysis | |
| τ²-Bench subset=banking | 3.9 % | independent | Artificial Analysis |
coding
| LiveCodeBench v6 implementation=artificial-analysis | 39.7 % | independent | Artificial Analysis | |
| SciCode | 33.1 % | independent | Artificial Analysis |
external indices
| AA Coding Index | 16.3 points | independent | Artificial Analysis | |
| Intelligence Index v4.1 | 14.3 points | independent | Artificial Analysis | |
| AA Math Index | 19.3 points | independent | Artificial Analysis |
factuality
| MASK | 49.7 % 48.1–51.3 | independent | Scale Labs |
knowledge science
| GPQA Diamond implementation=artificial-analysis | 67.1 % | independent | Artificial Analysis | |
| Humanity's Last Exam contamination=Potential contamination warning: This model was evaluated after the public release of HLE, allowing model builder access to the prompts and solutions. · implementation=scale | 5.7 % 4.8–6.6 | independent | Scale Labs | |
| Humanity's Last Exam implementation=artificial-analysis | 4.8 % | independent | Artificial Analysis | |
| MMLU-Pro implementation=artificial-analysis | 80.9 % | independent | Artificial Analysis |
long context instruction
| AA-LCR | 46.0 % | independent | Artificial Analysis | |
| IFBench | 43.0 % | independent | Artificial Analysis |
multimodal
| VISTA | 38.3 % 37.8–38.9 | independent | Scale Labs |
reasoning math
| AIME implementation=artificial-analysis | 39.0 % | independent | Artificial Analysis | |
| AIME year=2025 · implementation=artificial-analysis | 19.3 % | independent | Artificial Analysis | |
| EnigmaEval contamination=Potential contamination warning: This model was evaluated after the public release of EnigmaEval, allowing model builder access to the prompts and solutions. | 0.6 % 0.1–1.0 | independent | Scale Labs | |
| MATH-500 implementation=artificial-analysis | 88.9 % | independent | Artificial Analysis |
Agent + model results
systems, not bare-model scores
| agent + model Artificial Analysis harness + Llama 4 Maverick | Terminal-Bench 2.1 | 7.9 % | independent | Artificial Analysis |
| agent + model Artificial Analysis harness + Llama 4 Maverick | Terminal-Bench Hard | 6.8 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.