/models/anthropic/claude-opus-4-7.toml

name = "Claude Opus 4.7"
description = "Stronger Opus tier for advanced software work and high-stakes reasoning"
family = "claude-opus"
release_date = "2026-04-16"
last_updated = "2026-04-16"
attachment = true
reasoning = true
temperature = false
tool_call = true
knowledge = "2026-01-31"
open_weights = false

[limit]
context = 1_000_000
output = 128_000

[modalities]
input = ["text", "image", "pdf"]
output = ["text"]

[[benchmarks]]
name = "SWE-Bench Pro"
score = 64.3
metric = "resolve rate"
source = "https://www.anthropic.com/news/claude-opus-4-8"
date = "2026-05-28"

[[benchmarks]]
name = "Terminal-Bench"
score = 66.1
metric = "success rate"
harness = "Terminus-2"
version = "2.1"
source = "https://www.anthropic.com/news/claude-opus-4-8"
date = "2026-05-28"

[[benchmarks]]
name = "SWE-Atlas Refactoring"
score = 48.57
metric = "score"
harness = "Claude Code"
source = "https://labs.scale.com/leaderboard/sweatlas-refactoring"

[[benchmarks]]
name = "Artificial Analysis Coding Agent Index"
score = 66.6
metric = "average pass@1"
harness = "Claude Code"
variant = "max"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "SWE-Atlas Codebase QnA"
score = 81
metric = "pass@1"
harness = "Claude Code"
variant = "max"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "SWE-Bench Pro"
score = 44.9
metric = "pass@1"
harness = "Claude Code"
variant = "max"
dataset = "hard-aa"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "Terminal-Bench"
score = 73.8
metric = "pass@1"
harness = "Claude Code"
variant = "max"
version = "2.1"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "Artificial Analysis Coding Agent Index"
score = 61.2
metric = "average pass@1"
harness = "Cursor CLI"
variant = "medium"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "SWE-Atlas Codebase QnA"
score = 78.4
metric = "pass@1"
harness = "Cursor CLI"
variant = "medium"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "SWE-Bench Pro"
score = 34.4
metric = "pass@1"
harness = "Cursor CLI"
variant = "medium"
dataset = "hard-aa"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "Terminal-Bench"
score = 70.6
metric = "pass@1"
harness = "Cursor CLI"
variant = "medium"
version = "2.1"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "Artificial Analysis Coding Agent Index"
score = 59.9
metric = "average pass@1"
harness = "Claude Code"
variant = "medium"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "SWE-Atlas Codebase QnA"
score = 71.7
metric = "pass@1"
harness = "Claude Code"
variant = "medium"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "SWE-Bench Pro"
score = 36.4
metric = "pass@1"
harness = "Claude Code"
variant = "medium"
dataset = "hard-aa"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "Terminal-Bench"
score = 71.4
metric = "pass@1"
harness = "Claude Code"
variant = "medium"
version = "2.1"
source = "https://artificialanalysis.ai/agents/coding-agents"

[[benchmarks]]
name = "GPQA Diamond"
score = 94.2
metric = "accuracy"
source = "https://openai.com/index/introducing-gpt-5-5/"
date = "2026-04-23"

[[benchmarks]]
name = "Humanity's Last Exam"
score = 46.9
metric = "accuracy"
variant = "no tools"
source = "https://openai.com/index/introducing-gpt-5-5/"
date = "2026-04-23"

[[benchmarks]]
name = "Humanity's Last Exam"
score = 54.7
metric = "accuracy"
variant = "with tools"
source = "https://openai.com/index/introducing-gpt-5-5/"
date = "2026-04-23"

[[benchmarks]]
name = "OSWorld-Verified"
score = 78.0
metric = "success rate"
source = "https://openai.com/index/introducing-gpt-5-5/"
date = "2026-04-23"