AI Model Timeline

Tracking the accelerating release frequency of frontier AI models.

Claude Sonnet 4.5

Anthropic·

LLMsmidcloud

Overview

Balanced model with next-gen capabilities.

Capabilities and innovations

Real-time collaborationVoice modeAdvanced tool use

Benchmarks

BenchmarkScoreSource
GPQA Diamond81.6%unsourced
Humanity’s Last Exam (no tools)30.5%unsourced
SWE-bench Verified59.1%unsourced
MMLU89.8%unsourced
MMLU-Pro87.36%unsourced
  • SWE-bench Pro: not reported by the vendor
  • Terminal-Bench 2.x: benchmark did not exist at release
  • MMMU-Pro: benchmark did not exist at release

Links

More from Anthropic

Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.