Claude Opus 4.7

Released

Claude Opus 4.7 is an AI model released by Anthropic on Thursday, Apr 16 2026, 58 days after Claude Sonnet 4.6. It has a 1M token context window. Benchmark results (shown below) cover BullshitBench v2, ProgramBench, SWE-Bench Pro, SWE-Bench Verified, SWE-Bench Multilingual, Next.js Evals, and 17 more.

API pricing

Input
$5.00
Cached input
$0.50
5 min cache write
$6.25
1 hr cache write
$10.00
Output
$25.00
  • Reasoning or thinking is supported.
USD per 1M tokens

Available from

Amazon Bedrockglobal$5.00$25.001M
Azureglobal$5.00$25.001M
Googleglobal$5.00$25.001M
Amazon Bedrockeu-west-1$5.50$27.501M
Googleeu$5.50$27.501M
Googleus$5.50$27.501M
USD per 1M tokens

Benchmarks

Coding

ProgramBench
#5 of 17
SWE-Bench Pro
64.3%
#7 of 23
SWE-Bench Verified
87.6%
#5 of 57
SWE-Bench Multilingual
80.5%
#6 of 14
Next.js Evals
58%
#19 of 28

Terminal & CLI

Terminal-Bench 2.1
66.1%
#27 of 30
Terminal-Bench 2.0
69.4%
#5 of 14

Agentic & tool use

MCP Atlas
79.1%
#5 of 11
BrowseComp
79.3%
#20 of 28
CyberGym
73.1%
#10 of 10
OSWorld-Verified
78%
#11 of 28

Reasoning & science

Humanity's Last Exam
46.9%
no tools
54.7%
with tools
no tools
#4 of 22
with tools
#15 of 42
ARC-AGI-2
75.8%
#9 of 21
FrontierMath
43.8%
Tier 1–3
22.9%
Tier 4
Tier 1–3
#5 of 6
Tier 4
#5 of 6
GPQA Diamond
94.2%
#4 of 59

Knowledge work

GDPval-AA
1753
#4 of 9
GDPval (win/tie rate)
80.3%
#5 of 6

Finance

Finance Agent v2
51.5%
#6 of 9

Multimodal

CharXiv Reasoning
82.1%
#11 of 16
MMMU-Pro
75.2%
#10 of 12
Blueprint-Bench 2
24.5%
#4 of 6

Long context

MRCR v2 (8-needle)
59.3%
128k average
#7 of 7

Robustness

BullshitBench v2
83%
#6 of 77

Source: ProgramBench, retrieved 11 September 2026 · Source: BenchLM, retrieved 24 August 2026. Every other score here is the figure the lab published at launch.

Compare Claude Opus 4.7 with

Claude Opus 4.7

Suggested comparisons

Frequently asked questions

Claude Opus 4.7 was released by Anthropic on Thursday, Apr 16 2026.

All Anthropic releases

27 tracked