Claude Opus 5 Review: Features, Benchmarks & Pricing

Claude Opus 5 is Anthropic's Opus-class model released on July 24, 2026, built for complex, semi-autonomous professional work. It launched at the same $5/$25 price as Claude Opus 4.8, at half the cost of Claude Fable 5, and on several of Anthropic's benchmarks it matched or beat Fable 5 for a fraction of the price.

This Claude Opus 5 review covers what the model does well, what its benchmark numbers show, what it costs, how it compares with Claude Opus 5.5, which took over the top of the Opus line in September, and where its limits sit. It also covers access: Claude Opus 5 is not yet integrated into Chat Smith, but Claude Sonnet 5 and other Claude models already are.

Claude Opus 5 sits between Sonnet 5 and the Fable models in Anthropic's lineup. Sonnet 5 handles everyday work at $2 and $10 per million tokens, Opus 5 targets complex, multi-step tasks at $5 and $25, and Fable 5 covers the hardest problems at $10 and $50. Our Claude Fable 5 overview covers that top tier.

It supports a 1-million-token context window with up to 128,000 tokens of output, and at launch it became the default model on Claude Max and the strongest model available on Claude Pro. Anthropic also reported its lowest misaligned-behavior score of any recent Claude model in its pre-deployment audit.

Two months later, Anthropic released Claude Opus 5.5 at a lower price of $4 and $20, and it now leads the Opus line. Opus 5 remains available through the Claude API.

Four areas defined Opus 5 at launch.

Anthropic reported that on CursorBench 3.2, Opus 5 came within 0.5% of Fable 5's peak score at max effort for half the cost per task, and that on Frontier-Bench v0.1 it set a new state of the art at more than twice Opus 4.8's score for a lower cost per task. On Anthropic's later Terminal-Bench 4.0 runs, it scores 52.3%.

On OSWorld 2.0, which tests long computer-use workflows, Anthropic says Opus 5 beat Fable 5's best result at just over a third of the cost, with a score of 74.0%. That makes it strong at operating software, filling in forms, and moving between apps to finish a task.

On Zapier's AutomationBench, Anthropic reported roughly 1.5 times the next-best pass rate at the same cost per task when Opus 5 launched, along with strong results on GDPval-AA, Humanity's Last Exam, and DeepSearchQA. For document-heavy jobs such as drafting with an AI proposal generator, that mix of reasoning and follow-through is where Opus 5 shines.

On ARC-AGI 3, a test of reasoning on unfamiliar puzzles, Anthropic said Opus 5 scored three times higher than the next-best model at launch. It also scores 63.6% on Humanity's Last Exam in Anthropic's later comparison table.

The table below uses Anthropic's published figures on its current benchmark suite. Prices are API rates per million tokens.

The pattern is clear. Opus 5 sits well above Sonnet 5 on agentic coding and professional work, but Opus 5.5 beats it on every row while costing 20% less per token. For most new work on the API, Opus 5.5 is now the better Opus choice.

Claude Opus 5 costs $5 per million input tokens and $25 per million output tokens, the same as Opus 4.8 and half of Fable 5. Batch processing halves that to $2.50 and $12.50, cache reads cost $0.50 per million tokens, and Fast mode runs about 2.5 times faster at $10 and $50.

• Claude apps: available in Claude.ai, where it launched as the default model on Claude Max and the strongest model on Claude Pro.
• Developers: available as claude-opus-5 on the Claude Platform and major cloud platforms.
• Chat Smith: Claude Opus 5 is not integrated yet. You can use Claude Sonnet 5 and other Claude models today, with higher usage limits on Chat Smith Pro at $9.99 per month or $39.99 per year.

Opus 5 has already been superseded. Claude Opus 5.5 arrived on September 22, 2026 with better scores across Anthropic's benchmark table, more than 30% faster output, and a 20% lower price per token.

It is also slow and expensive for everyday work. Independent tracking puts its output at around 56 tokens per second, with a long wait before the first token at high effort, and at $25 per million output tokens it costs 2.5 times as much as Sonnet 5 for routine tasks that do not need Opus-level reasoning.

As always, the published benchmark figures are mostly Anthropic's own. Test Opus 5 against alternatives such as DeepSeek V4 Pro on your real tasks before committing.

Claude Opus 5 is not integrated into Chat Smith yet, but several Claude models already are. Pick one from the model list and start chatting on Android, iOS, or in the browser, with no Anthropic account needed.

Claude Sonnet 5 is the closest option for most work: strong writing, research, and coding at a fraction of Opus pricing, with a 1M-token context window. Our Claude Sonnet 5 review covers its strengths in detail. For quick tasks, Claude Haiku 4.5 is faster, and Claude Sonnet 4.6 remains available if you have prompts tuned for it. Opus 5's biggest advantages show up mainly in long agentic coding, computer use, and complex business automation.

Because Chat Smith also includes models from OpenAI, Google, xAI, and DeepSeek, you can put the same prompt to a Claude model and its rivals and keep the answer that holds up. Chat Smith Pro adds higher usage limits and more premium models for $9.99 per month or $39.99 per year.

Claude Opus 5 was a major step for Anthropic's Opus line: near-Fable results on coding and computer use at half Fable's price, strong business-automation and reasoning scores, and a 1M-token context window. Today, its main drawback is that Opus 5.5 does more for less.

If you want Claude's strengths without API costs, the Claude models already in Chat Smith are an easy way in. You can try them now in Chat Smith, an AI chatbot that puts Claude next to the latest models from OpenAI, Google, xAI, and DeepSeek.