Frontier BenchAI model benchmarks · Jul 2026
ModelsCompareLabsMethodology
All models
A

Claude Opus 4.8

Anthropic

Current agentic-coding and computer-use leader. Best default for engineering teams.

FrontierProprietary1M contextReleased May 2026textvisioncode

API price · per 1M tokens

$5/$25

input / output

≈ $10.00/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.
88.6/ best 90.1
SWE-bench Pro2,294 harder real-world GitHub issues. Where the frontier separates now.
69.2/ best 69.2

Best in class

Terminal-Bench 2.xAgentic command-line workflows — planning, iteration, tool coordination.
74.6/ best 84.0
GPQA DiamondGoogle-proof graduate-level science Q&A. General reasoning proxy.
93.6/ best 94.4
MMLU-ProMulti-task language understanding across 120+ academic subjects.
84.8/ best 93.0
HumanEval+Functional code generation from docstrings.
94.1/ best 96.0
MATH-500Competition-style math problems.
91.4/ best 97.8
Humanity's Last ExamHardest reasoning benchmark (with tools). ~50% = near frontier.
57.9/ best 57.9

Best in class

AIME 2025American Invitational Mathematics Examination.
92.0/ best 96.5
OSWorldReal computer-use tasks across desktop apps.
83.4/ best 83.4

Best in class

Other Anthropic models

A

Claude Opus 4.7

Frontier

≈$10.00/MA

Claude Sonnet 4.7

Mid

≈$6.00/MA

Claude Haiku 4.7

Fast / Cheap

≈$1.60/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.