Frontier Log

AnthropicModel

Claude Opus 4.5

Made by
Anthropic
Released
Announced in
Introducing Claude Opus 4.5anthropic.com
Event
Claude Opus 4.5
API record
/api/v1/models/claude-opus-4-5

Benchmark results

BenchmarkScoreSource check
Humanity's Last ExamNo tools · reasoning only; Claude Sonnet 4.5 as grader30.8%
CheckedA person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 2 other lab documents: 1 2
Humanity's Last ExamWith tools · web search, web fetch and code execution, no reasoning; Claude Sonnet 4.5 as grader43.2%
CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 2 other lab documents: 1 2
web search, web fetch, code execution, programmatic tool calling; compaction at 50K up to 3M tokens; max effort, adaptive thinking · reported later, in Claude Opus 4.643.4%
CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 1 other lab document: 1
SWE-bench Verifiedno thinking budget, high effort; averaged over 5 trials80.9%
CheckedA person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 8 other lab documents, the first 3 linked: 1 2 3
SWE-bench Prono thinking; average of 5 trials · The source doesn't say whether this is the public or the commercial set52.0%
CheckedA person checked this number against its source on 19 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
64K thinking budget; average of 5 trials · The source doesn't say whether this is the public or the commercial set51.60%
CheckedA person checked this number against its source on 19 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
GPQA Diamond64K thinking budget, high effort; averaged over 5 trials87.0%
CheckedA person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 8 other lab documents, the first 3 linked: 1 2 3
BrowseCompreported later, in Claude Opus 4.667.8%
CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 3 other lab documents: 1 2 3
ARC-AGI-264K thinking budget, high effort; averaged over 5 trials37.6%
CheckedA person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 3 other lab documents: 1 2 3
Terminal-Bench 2.0128K thinking budget, high effort; averaged over 5 trials59.3%
CheckedA person checked this number against its source on 16 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 4 other lab documents, the first 3 linked: 1 2 3
reported later, in Claude Opus 4.6 System Card59.8%
CorroboratedThe same number is printed in 3 other lab documents: 1 2 3
Sources

Reported by Anthropic

Events

Stories

All models