Publication date: July 27, 2026
Anthropic released Claude Opus 5, the new flagship of its Claude model family. Claude Opus 5 ships with a 1 million token context window, scores 96.0% on SWE-bench Verified, and keeps Opus-tier pricing at $5 per million input tokens and $25 per million output tokens. The model took the top spot on the Artificial Analysis Intelligence Index at launch and is Anthropic’s fourth model in under two months.
What is Claude Opus 5 and what can it do?
Claude Opus 5 is Anthropic’s most capable model, positioned above Claude Sonnet 5 and Claude Fable 5 in the Claude 5 family. It is aimed at agentic coding, computer use, and long-horizon reasoning tasks. Extended thinking is on by default, and a per-request effort toggle with low, medium, and high settings lets developers raise reasoning compute for hard problems or lower it for routine work.
The model handles a 1 million token context window as both default and maximum, with a standard maximum output of 128,000 tokens. That context size lets Claude Opus 5 work across large codebases, long documents, and multi-step agent runs without losing earlier context. Anthropic used the pre-launch codename Honeycomb during development.
Claude Opus 5 benchmarks and technical specs
Claude Opus 5 leads the Artificial Analysis Intelligence Index at 61 and the Agentic Index at 55.3, placing it ahead of Claude Fable 5 (60 and 52.8) and OpenAI’s GPT-5.6 Sol (59 and 54.0) on the combined intelligence measure at launch.
On coding, Claude Opus 5 scored 96.0% on SWE-bench Verified and 79.2% on SWE-bench Pro. On broader reasoning tests it reached 43.3% on Frontier-Bench and 30.2% on ARC-AGI-3, where the next-best model scored 7.8%. For computer use, Claude Opus 5 reached 70.57% on OSWorld 2.0, up from 55.7% for its predecessor Claude Opus 4.8.
How does Claude Opus 5 compare to Fable 5 and GPT-5.6 Sol?
Against Anthropic’s own Claude Fable 5, Claude Opus 5 leads on overall intelligence and agentic scores and matches it closely on coding, though Fable 5 still edges Opus 5 on SWE-bench Pro at 80.0% versus 79.2%. Claude Opus 5 also costs half as much as Fable 5, which is priced at $10 per million input tokens and $50 per million output tokens.
Against OpenAI’s GPT-5.6 Sol, Claude Opus 5 ranks higher on both the Artificial Analysis Intelligence Index (61 versus 59) and the Agentic Index (55.3 versus 54.0). On the ARC-AGI-3 test for novel reasoning, Claude Opus 5 posts a large lead, at 30.2% against 7.8% for the next-best model.
Claude Opus 5 pricing and availability
Anthropic priced Claude Opus 5 at $5 per million input tokens and $25 per million output tokens, unchanged from Claude Opus 4.8 and half the price of Claude Fable 5. Extended thinking and the effort toggle are included at the same token price, so the reasoning controls do not carry a separate fee.
The model is available through the Anthropic API and the major cloud platforms that carry the Claude family, including providers such as Amazon Bedrock. Developers already using Claude Opus 4.8 can move to Claude Opus 5 without a pricing change.
Claude Opus 5 is Anthropic’s new flagship model, released on July 24, 2026, with a 1 million token context window, a 96.0% SWE-bench Verified score, and pricing held at $5 and $25 per million tokens.
Full details are in Anthropic’s official announcement at anthropic.com/news/claude-opus-5.
