Claude Opus 4.7
anthropic-claude-opus-4-7
fields
| nameREQ | Claude Opus 4.7 | www.anthropic.com | 2026-04-30 |
| vendorREQ | Anthropic | www.anthropic.com | 2026-04-30 |
| released_at | 2026-04-16 | www.anthropic.com | 2026-04-30 |
| modality | text, code, vision | www.anthropic.com | 2026-04-30 |
| access | closed | www.anthropic.com | 2026-04-30 |
| context_window | |||
| parameters | |||
| benchmark_score | 64.32 revisions | www.buildfastwithai.com | 2026-05-03 |
| benchmark_name | SWE-Bench Pro2 revisions | www.buildfastwithai.com | 2026-05-03 |
| announcement_url | https://www.anthropic.com/news/claude-opus-4-7 | www.anthropic.com | 2026-04-30 |
history · 8 fields · 10 revisions
name1 revision
Claude Opus 4.7
current
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.
vendor1 revision
Anthropic
current
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.
released_at1 revision
2026-04-16
current
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.
modality1 revision
text, code, vision
current
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.
access1 revision
closed
current
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.
benchmark_score2 revisions
64.3
current
Claude Opus 4.7 pushed SWE-bench Pro from 53.4% to 64.3%, reclaiming the coding crown. Claude Opus 4.7 leads SWE-bench Pro at 64.3% for real GitHub issue resolution.
69.4
superseded
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.
benchmark_name2 revisions
SWE-Bench Pro
current
Claude Opus 4.7 pushed SWE-bench Pro from 53.4% to 64.3%, reclaiming the coding crown.
Terminal-Bench 2.0
superseded
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.
announcement_url1 revision
https://www.anthropic.com/news/claude-opus-4-7
current
Claude Opus 4.7: notable improvement on Opus 4.6 in advanced software engineering. Accepts images up to 2,576 pixels. $5/M input, $25/M output tokens.