Frontier BenchAI model benchmarks · Jul 2026
ModelsCompareLabsMethodology
All models
MA

Llama 4 Maverick

Meta AI

MoE flagship with deep ecosystem tooling. 17B active, 400B total.

Open-weightLlama license1M contextReleased Jan 2026textvisioncode

API price · per 1M tokens

$0.2/$0.6

input / output

≈ $0.30/M blended

Benchmarks

Bar shows this model's score; tick shows the best score across all tracked models for context.

SWE-bench VerifiedReal GitHub issues, human-verified patches. The standard coding-agent benchmark.
70.0/ best 90.1
SWE-bench Pro2,294 harder real-world GitHub issues. Where the frontier separates now.
44.0/ best 69.2
GPQA DiamondGoogle-proof graduate-level science Q&A. General reasoning proxy.
80.0/ best 94.4
MMLU-ProMulti-task language understanding across 120+ academic subjects.
80.5/ best 93.0
HumanEval+Functional code generation from docstrings.
88.0/ best 96.0
MATH-500Competition-style math problems.
86.0/ best 97.8

Other Meta AI models

MA

Llama 4 Scout

Open-weight

≈$0.22/M

Built by GLM-5.2 · Scores are directional, drawn from public mid-2026 benchmarks.

SWE-bench, GPQA, MMLU, AIME, HLE etc. are trademarks of their respective owners.