← Back to directory
Anthropic
Claude Opus 4.1
Released Aug 5, 2025 · knowledge cutoff 2025-03
- Status
- Superseded
- Location
- United States
- Modality
- Multimodal
- Context window
- 200K
- Max output
- 32K
- Speed
- —
- Price ($/MTok in / out)
- $15 / $75
- Cost per task
- —
- Open weights
- No
Benchmarks4 of 8 reported
GPQA Diamond
80.9%
SWE-bench Verified
74.5%
AIME (latest)
78%
LMArena Elo
1430
MMLU-Pro
—
Terminal-Bench 2.1
—
Humanity's Last Exam
—
ARC-AGI-2
—
Strengths & weaknesses
Strengths
- Holds several files in mind at once, so a refactor stays consistent across all of them
- Fewer collateral regressions — less likely to break something adjacent while fixing a bug
- Behaviourally a drop-in: existing prompts kept working unchanged
Weaknesses
- Feels like the same model with sharper edges rather than a new one
- Same slow, deliberate pacing as Opus 4
For developers
API model strings
Not researched
Licence
Proprietary — weights not released
Retirement
No retirement announced
Lineage
No recorded predecessor