Opus 4.7 is now available in Kiro

Starting today, Claude Opus 4.7 is rolling out in the Kiro IDE and CLI. Opus 4.7 is Anthropic's latest and most capable Opus model, a direct upgrade from Opus 4.6, with stronger coding performance on complex, long-running tasks.

Comparison table of AI model performance across 14 benchmarks for five models: Opus 4.7, Opus 4.6, GPT-5.4, Gemini 3.1 Pro, and Mythos Preview. Agentic coding on SWE-bench Pro: Opus 4.7 64.3%, Opus 4.6 53.4%, GPT-5.4 57.7%, Gemini 3.1 Pro 54.2%, Mythos Preview 77.8%. Agentic coding on SWE-bench Verified: Opus 4.7 87.6%, Opus 4.6 80.8%, GPT-5.4 N/A, Gemini 3.1 Pro 80.6%, Mythos Preview 93.9%. Agentic terminal coding (Terminal Bench 2.0): Opus 4.7 69.4%, Opus 4.6 65.4%, GPT-5.4 75.1%, Gemini 3.1 Pro 68.5%, Mythos Preview 82.0%. Multidisciplinary reasoning (Humanity's Last Exam) no tools: Opus 4.7 46.9%, Opus 4.6 40.0%, GPT-5.4 42.7%, Gemini 3.1 Pro 44.4%, Mythos Preview 56.8%. Multidisciplinary reasoning with tools: Opus 4.7 54.7%, Opus 4.6 53.3%, GPT-5.4 58.7%, Gemini 3.1 Pro 51.4%, Mythos Preview 64.7%. Agentic search (BrowseComp): Opus 4.7 79.3%, Opus 4.6 83.7%, GPT-5.4 89.3%, Gemini 3.1 Pro 85.9%, Mythos Preview 86.9%. Scaled tool use (MCP-Atlas): Opus 4.7 77.3%, Opus 4.6 75.8%, GPT-5.4 68.1%, Gemini 3.1 Pro 73.9%, Mythos Preview N/A. Agentic computer use (OSWorld-Verified): Opus 4.7 78.0%, Opus 4.6 72.7%, GPT-5.4 75.0%, Gemini 3.1 Pro N/A, Mythos Preview 79.6%. Agentic financial analysis (Finance Agent v11): Opus 4.7 64.4%, Opus 4.6 60.1%, GPT-5.4 61.5%, Gemini 3.1 Pro 59.7%, Mythos Preview N/A. Cybersecurity vulnerability reproduction (CyberGym): Opus 4.7 73.1%, Opus 4.6 73.8%, GPT-5.4 66.3%, Gemini 3.1 Pro N/A, Mythos Preview 83.1%. Graduate-level reasoning (GPQA Diamond): Opus 4.7 94.2%, Opus 4.6 91.3%, GPT-5.4 94.4%, Gemini 3.1 Pro 94.3%, Mythos Preview 94.6%. Visual reasoning (CharXiv Reasoning) no tools: Opus 4.7 82.1%, Opus 4.6 69.1%, GPT-5.4 N/A, Gemini 3.1 Pro N/A, Mythos Preview 86.1%. Visual reasoning with tools: Opus 4.7 91.0%, Opus 4.6 84.7%, GPT-5.4 N/A, Gemini 3.1 Pro N/A, Mythos Preview 93.2%. Multilingual Q&A (MMLU): Opus 4.7 91.5%, Opus 4.6 91.1%, GPT-5.4 N/A, Gemini 3.1 Pro 92.6%, Mythos Preview N/A. Opus 4.7 is highlighted with an orange border. Mythos Preview leads in most benchmarks where tested; graduate-level reasoning scores are consistently above 91% across all models.

Opus 4.7 resolves more production tasks than its predecessor and follows complex instructions more precisely across longer sessions. It handles multi-step implementations that span multiple files and tools, verifies its own outputs before returning results, and holds closer to what you asked for with stronger follow-through.

Those improvements carry into Kiro's spec-driven development. We see Opus 4.7 as the best model fit for carrying detailed specs into implementation with higher fidelity across larger codebases and broader changes. In workflows that move between planning, tool use, execution, and review, it holds its thread with less drift.

Opus 4.7 is rolling out gradually with experimental support to a subset of Kiro Pro, Pro+, and Power customers logging in with AWS IAM Identity Center in the AWS US-East-1 (Northern Virginia) and AWS Europe (Frankfurt) regions, with cross-region inference support and broader availability to follow. It ships with the full 1M context window, 2.2x credit multiplier, the same as Opus 4.6.

Download Kiro or restart the app or CLI to check for the latest available models.

Root cause in 33 seconds: How Kiro CLI saved 4 years of build time