Claude Opus 4.7

Anthropic

flagship

Capabilities

VisionToolsStructured OutputExtended ThinkingCodeLong ContextMultilingualComputer UseVideoAudio

Strengths

Excellent long-document summarisation (1M context window)
Strong vision and multimodal understanding
Superior agentic and tool-use capabilities
Structured output and extended thinking support

Weaknesses

High output pricing ($25/M tokens) increases per-task cost
Slower inference due to large model size
Closed-source limits transparency and self-hosting
Extended thinking adds latency for reasoning tasks

Pricing

Input / 1M tokens

Output / 1M tokens

$25

Context window

1,000k

Transparency

Open weights

0.0 / 10

Open training data

0.0 / 10

Open methodology

3.5 / 10

Licence openness

2.0 / 10

Provider disclosure

4.5 / 10

FMTI company score

N/A

Composite:3.5 / 10

Closed-source model with limited public disclosure on training data and methodology. Anthropic publishes some research on Constitutional AI and alignment but does not open-source weights or full training details.

Sustainability

Inference energy

N/A

Training footprint

N/A

Provider infrastructure

5.5 / 10

Composite:5.0 / 10

Large flagship model with high computational cost per inference. Anthropic's sustainability commitments are not prominently published; infrastructure details unavailable. Assume moderate environmental footprint typical of large closed-source providers.

Task Fitness

question answering

90%

code

90%

mathematics

91%

business communication

75%

analysis

91%

extraction

94%

long-form generation

88%

research

64%

reasoning

91%

summarisation

92%

translation

87%

conversation

89%