Claude Opus 5 Launches With 1M Context at Half the Price of Fable 5

Anthropic has added Claude Opus 5 to its model lineup. Its API ID is claude-opus-5, and its target use case is clear: complex agentic coding and enterprise-grade work.

The most important detail is not parameter count. Anthropic still has not disclosed the model’s weight size. The real change is that Opus 5 gets a 1M-token context window, 128K maximum output, and default adaptive thinking, while staying in the Opus price band: $5 input and $25 output per million tokens.

In one line: Opus 5 brings stronger agent capability down into the Opus price tier instead of creating another more expensive Fable-class model.

30-Second Summary

• Model ID: claude-opus-5
• Positioning: complex agentic coding, enterprise knowledge work, multi-step tasks
• Input modalities: text and image
• Output modality: text
• Context window: 1M tokens
• Maximum output: 128K tokens
• Batch API maximum output: up to 300K tokens with a beta header
• Thinking: adaptive thinking, enabled by default
• Default effort: high
• Knowledge cutoff: reliable knowledge and training data cutoffs are both May 2026
• Parameter count: not publicly disclosed

Pricing: Opus 5 Does Not Move Into the Fable Tier

Anthropic’s official pricing lists Claude Opus 5 at:

Billing item	Claude Opus 5
Standard input	$5 / MTok
5-minute cache write	$6.25 / MTok
1-hour cache write	$10 / MTok
Cache hit / refresh	$0.50 / MTok
Output	$25 / MTok

That keeps Opus 5 at the same base price as Opus 4.8, while making it half the price of Claude Fable 5. Fable 5 is $10 input / $50 output; Opus 5 is $5 input / $25 output.

For a rough workload of 1 million input tokens plus 200,000 output tokens:

Model	Estimated cost
Claude Opus 5	$10
Claude Fable 5	$20
Claude Sonnet 5 promotional price	$4
GPT-5.6 Sol short-context price	$11

That puts Opus 5 close to GPT-5.6 Sol’s short-context standard pricing. Inside the Claude family, it looks much more like a daily high-end workhorse than a Fable 5 replacement.

How to Choose Across the Claude Family

Model	Positioning	Input / output price	Context	Max output	Best fit
Claude Fable 5	Strongest broadly available Claude model, long-running agents	$10 / $50	1M	128K	Highest quality, long tasks, complex reasoning
Claude Opus 5	Complex agentic coding and enterprise work	$5 / $25	1M	128K	High-quality coding, enterprise knowledge work, long-running agents
Claude Sonnet 5	Balance of speed and intelligence	$2 / $10 through August 31	1M	128K	High-throughput and cost-sensitive production tasks
Claude Haiku 4.5	Fastest and lowest-cost option	$1 / $5	200K	64K	Classification, extraction, lightweight automation

Anthropic’s guidance is direct: if you are unsure where to start for complex agentic coding or enterprise work, start with Claude Opus 5. If you need the strongest available capability, use Claude Fable 5.

That creates a practical selection rule:

• Highest quality: Fable 5
• Daily complex agents and coding tasks: Opus 5
• More cost-sensitive speed: Sonnet 5
• Large-scale lightweight tasks: Haiku 4.5

What Changes From Opus 4.8?

Opus 5 is not just a name bump. Several changes affect day-to-day developer experience.

1. Thinking Is On by Default

Opus 5 uses adaptive thinking, adjusting its depth of reasoning based on the task. For multi-file code changes, reviews, and complex document analysis, it behaves more like a model that actively allocates reasoning budget.

2. It Is Better Suited to Long-Running Agents

The 1M context window and 128K output limit let Opus 5 handle larger repositories, longer evidence packs, and more complete task trajectories. The Batch API’s 300K output ceiling also signals room for very long asynchronous jobs.

3. Pricing Does Not Climb With Capability

This is the commercial signal that matters. Fable 5 is the top capability tier. Opus 5 looks more like the high-capability daily tier. For teams, that means more complex work can move to Opus 5 without sending every task to the most expensive model.

Why the Parameter Count Should Stay Blank

The obvious question is: how many parameters does Opus 5 have?

For now, the answer is simple: Anthropic has not publicly disclosed it.

So articles, comparisons, and model tables should avoid invented trillion-parameter claims. Claude releases have historically emphasized capability, context length, output limits, pricing, and model behavior rather than public weight size.

For users, the verifiable selection data is:

• API ID
• context length
• maximum output
• input and output pricing
• thinking and effort behavior
• official recommended use cases

Those are the numbers that actually map to cost and product experience.

Conclusion

Claude Opus 5 is not framed as stronger than Fable 5. Its real move is putting complex agentic and enterprise capability into a more sustainable price tier.

It keeps 1M context and 128K output while costing half as much as Fable 5. For many developers and teams, Opus 5 may become the default high-end Claude model: stronger than Sonnet 5 for hard tasks, but much easier to use continuously than Fable 5.

Sources:

• Claude Models Overview
• Claude Pricing
• OpenAI API Pricing