Opus 5 changes which Claude model most people should try first
Anthropic released Claude Opus 5 on July 24, 2026. It is now the strongest model included with Claude Pro and the default model on Claude Max.
Anthropic says Opus 5 approaches Fable 5 on difficult work at roughly half the API price. That makes Opus the sensible first test for research, planning, coding, and business work before you reach for the more expensive model.
The benchmark chart is evidence, not a universal winner
Anthropic's comparison shows Opus 5 ahead of Fable 5 on several business-workflow, tool-use, and problem-solving evaluations. Other rows favor Fable, and a benchmark cannot tell you how many corrections your real task will need.
The useful conclusion is not that Fable is dead. It is that paying more or choosing the highest model name no longer settles the decision.
Start with Opus for difficult everyday work
Use Opus 5 first when you are researching a decision, planning a launch, reviewing documents, writing or debugging code, or working across a normal business workflow.
Opus is available inside the $20 Claude Pro plan in the United States. Plan pricing, usage limits, and model access can vary by country, workspace, and current product rules.
Keep Fable for the jobs that actually need it
Anthropic still positions Fable 5 for some specialized work and projects that may run autonomously for days. That is a different job from asking Claude to analyze a report, plan a campaign, or help with a normal coding session.
If the task is unusually long-running, highly autonomous, or specialized, test Fable. Otherwise, make Opus prove insufficient before moving up.
Run one fair comparison before you choose
- Pick one task you repeat often enough to matter.
- Give both models the same prompt, files, context, and definition of done.
- Count the corrections needed before each answer becomes usable.
- Compare response time and, for API work, total token cost rather than the headline rate alone.
- Keep the cheaper or more accessible model when the final quality is effectively the same.
Half the API price does not mean half off Claude Pro
Anthropic's price comparison refers to API rates. It is not a 50 percent discount on a Claude subscription, and it does not promise that Opus will use the same number of tokens or perform equally on every task.
For subscription users, the immediate change is simpler: Opus 5 is already included with Pro, so try it on the hard task before assuming you need a higher-priced option.