Anthropic Opus 5 Cuts AI Model Price With Longer-Answer Risk

Anthropic has released Opus 5 with a lower token price than Fable 5, but the cheaper list price does not settle what enterprise AI users will pay for completed work.

The Register article put Fable 5 input tokens at $10 / MTok and output tokens at $50 / MTok, while Opus 5 was priced at $5 / MTok and $25 / MTok.

The price cut arrives with two operating limits for buyers.

Model tasks consume different volumes of tokens, and Anthropic warns customers that Opus 5 may produce longer replies in default user-facing settings unless prompts ask for shorter output.

Artificial Analysis Task Costs Narrow The Gap

Artificial Analysis reported a weighted average cost per Intelligence Index task of $2.03 for Opus 5 max, compared with $2.75 for Fable, $0.95 for Kimi K3 and $1.04 for GPT-5.6 Sol max; the Intelligence Index ranked Opus 5 at 61, one point ahead of Fable.

The task-level comparison gives buyers a procurement view beyond headline token prices.

A company choosing between frontier models still has to consider output length, task mix and whether a model needs more tokens to reach the same result.

Prompt Changes Affect Enterprise Controls

Anthropic also changed the prompt environment around its latest models.

Claude Code engineer Thariq Shihipar wrote that the latest models cut the Claude Code system prompt by 80 percent, and Anthropic's guidance now tells users to simplify system prompts, skills and CLAUDE.md files.

The guidance also points customers toward a command called claude doctor for automatic optimisation of some prompts and skills.

Enterprise rollout therefore includes prompt-maintenance work across internal coding workflows, documentation and automation scripts.

Cyber Controls Remain Model-Specific

The system card assigns Opus 5 a different cybersecurity profile from Anthropic's other models.

Anthropic's system card recorded an OSS-Fuzz score of 80 percent for Opus 5 on finding open-source code security flaws, compared with 79.4 percent for Mythos, while Opus 5 succeeded in 4/14 exploitation attempts compared with 13/14 for Mythos.

Anthropic's cyber controls also separate vulnerability discovery from higher-risk activity.

The company says Opus 5 cyber classifiers are less restrictive than those on Fable 5 for finding source-code vulnerabilities, while blocking binary-based vulnerability scanning, penetration testing and exploit generation.

An automated fallback function is being added to the API so requests judged too risky for Opus 5 or Fable 5 can be routed to a lower-capability model rather than ending as outright blocks.

The safety layer therefore affects which model handles a request, not only whether the request fails.

Data Retention Becomes A Selling Point

The Register framed the lack of a data-retention requirement as a likely business advantage for Opus 5.

For companies sending code, prompts or internal workflow material into a hosted model, retention rules can matter as much as price when legal, security and procurement teams review deployment.

Anthropic has not published customer adoption numbers for Opus 5 or independent enterprise cost studies linked to the new model.

The public record also does not yet show whether longer default answers offset the lower token price in production workloads.