Claude Opus 5

• 🏢 Anthropic's most capable Opus-tier model, released July 2026.
• 📏 One-million-token context window with 128K max output tokens.
• 🧠 Adaptive thinking for long-horizon reasoning.
• 👁️ Multimodal vision plus coding, function calling, and web search.
• 🔧 Optimized for agentic coding and professional knowledge work.
• 🎯 Anthropic reports gains over Opus 4.8 across knowledge work.
• ⚡ Paired with a faster variant, Claude Opus 5 Fast.
• 🆕 New flagship above the Opus 4.x line.

Anthropic

Anthropic PBC is an American artificial intelligence company headquartered in San Francisco. Structured as a public benefit corporation, the lab develops large language models under the Claude name, with a research emphasis on building reliable, steerable,…

Claude Opus 5 is the flagship of Anthropic's Opus tier, released on July 23, 2026, and positioned as a generational upgrade for long-running agents, coding, and professional knowledge work over its predecessor, Claude Opus 4.8. It carries a one-million-token context window and 128K maximum output tokens, adds adaptive thinking for long-horizon reasoning, and retains strong multimodal vision and tool-use capabilities, including function calling and web search.

Anthropic frames Opus 5 as a step-change over Opus 4.8 rather than a raw-frontier leap, reporting improvements in agentic coding, professional knowledge work, and reasoning. A companion release, Claude Opus 5 Fast, targets latency-sensitive and cost-conscious workloads, giving deployers a lighter option within the same generation.

Opus 5 sits alongside recent siblings in the Claude 5 lineup, including Claude Sonnet 5 in the mid tier and Claude Fable 5, and it continues Anthropic's cadence of iterative Opus releases from Opus 4.5 through 4.8. For teams already building on earlier Opus models, it is a drop-in successor that keeps the same broad capability profile while extending context length and, per Anthropic, sharpening performance on demanding agentic and knowledge-work tasks.