TLDR
Anthropic has released Claude Opus 5 at $5 per million input tokens and $25 per million output tokens, matching the price of its predecessor Opus 4.8 and coming in at roughly half the cost of flagship Fable 5. Independent benchmarker Artificial Analysis ranked Opus 5 first on both its general Intelligence Index, scoring 61 against Fable 5's 60 and GPT-5.6 Sol's 59, and its Agentic Index with a top score of 55. On coding benchmarks, Opus 5 came within 0.5 per cent of Fable 5's peak score at approximately half the cost per task. The release sharpens the pricing war between Anthropic, OpenAI and Google at a moment when near-flagship AI capability is becoming a mid-tier commodity.
KEY TAKEAWAYS
What Anthropic released and what it costs
Anthropic released Claude Opus 5 on 24 July 2026, positioning the model as a strategic midpoint between its cost-efficient Sonnet 5 tier and the frontier Fable 5 flagship.[1] Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens, identical to its predecessor Opus 4.8 and roughly half the cost of Fable 5 per task.verifiedVerified Source: anthropic.com[1]
Anthropic made Opus 5 the default model on Claude Max and the strongest model available on Claude Pro at launch.[1] That decision puts a model benchmarked above the previous flagship directly into the hands of the platform's most active users, at no extra cost.
How Opus 5 benchmarks against Fable 5 and GPT-5.6 Sol
Artificial Analysis, which runs independent model evaluations, ranked Opus 5 first on its general Intelligence Index with a score of 61, ahead of Fable 5 at 60 and GPT-5.6 Sol at 59.[2] The margin is narrow, but the ordering holds across both general and agentic domains.
On the Artificial Analysis Agentic Index, Opus 5 scored 55 at max effort, the highest figure among all models evaluated.verifiedVerified Source: artificialanalysis.ai[3] Agentic benchmarks test a model's capacity to execute multi-step tasks autonomously, which is where enterprise AI spending is concentrated right now.
On CursorBench 3.2, the coding benchmark maintained by the Cursor development environment, Opus 5 came within 0.5 per cent of Fable 5's peak score at max effort while costing roughly half as much per task.verifiedVerified Source: anthropic.com[1] Sualeh Asif, co-founder of Cursor, said: "Claude Opus 5 delivers near Fable 5 intelligence at Opus speed and cost. On CursorBench it's just under Fable 5 and has many of the same behaviours. We are excited to see how developers use it in Cursor."[1]
Where Opus 5 is strongest
Scott Wu, CEO of Devin developer Cognition, said: "On FrontierCode 1.1, Claude Opus 5 approaches Fable-level performance at half the cost. Within Devin, it also shows particular strength on difficult debugging and root-cause analysis tasks."[1] That points to two capability areas that matter most for engineering teams: raw code generation and the slower, harder work of diagnosing why something broke.
Anthropic's own evaluation on Frontier-Bench v0.1 found Opus 5 surpasses all other models on that suite, delivering more than twice Opus 4.8's performance at a lower cost per task.[1] Anthropic cites agentic pipelines, terminal automation and knowledge work as primary use cases, reflecting where the model's Agentic Index score translates into practical deployment.
What near-flagship capability at mid-tier pricing means for businesses
For teams running AI-driven software engineering or data analysis at scale, the cost structure of Opus 5 changes the maths on deployment.[1] At $5 per million input tokens, organisations that previously reserved Fable 5 for high-value tasks can now route a wider share of workloads through a model that benchmarks above the previous generation's frontier.
Anthropic's lineup now spans Sonnet 5 for cost-efficient agentic tasks, Opus 5 as the mid-tier workhorse, and Fable 5 at the frontier premium.[1] Businesses can design tiered workflows where Fable 5 handles the narrow slice of tasks where its 0.5-per-cent coding advantage justifies the extra cost, and Opus 5 absorbs the rest.
The broader pricing war and what comes next
Artificial Analysis said that Opus 5 cutting cost per task by around 26 per cent relative to Fable 5 reflects a broader market shift in which top-tier capabilities are becoming accessible at progressively lower price points.[2] Anthropic, OpenAI and Google are all competing on this curve simultaneously, compressing the premium that frontier models can sustain.
Opus 5's arrival at the top position on both the Artificial Analysis Intelligence Index and the Agentic Index as of 24 July 2026 gives Anthropic a technical headline to defend before competitors respond.[2] The Agentic Index score of 55 is the figure to watch: agentic performance is where enterprise contract values are decided, and a lead there carries more commercial weight than general benchmark rankings alone.
SOURCES & CITATIONS
FREQUENTLY ASKED QUESTIONS
How much does Claude Opus 5 cost?
How does Opus 5 compare to Fable 5 and GPT-5.6 Sol on benchmarks?
Where is Claude Opus 5 available?
What tasks is Opus 5 best suited for?
Jonas Valenti writes about search and how businesses get discovered. He has spent years watching what makes a company visible online, and is unsentimental about tactics that no longer work.
