How to Keep Claude Fable 5 Costs Under Control | Upstash Blog<br>Login

On this pageWhere the money goes<br>Tier your models: Fable to think, cheaper models to do<br>Match effort to the task<br>Keep prompt caching intact<br>Trim the context that keeps getting re-read<br>Write skills instead of making the agent explore<br>Cut input tokens at the source with Context7<br>Putting it together

Claude Fable 5 is back. Anthropic pulled it on June 12, 2026, three days after announcing it, after Amazon researchers found a jailbreak that got the model to identify software vulnerabilities and even demonstrate an exploit, which triggered a U.S. export-control directive suspending access. It returned on July 1 with a new safety classifier. Fable is Anthropic's most capable model, and it is priced like it.

Per Anthropic's pricing, Fable costs $10 per million input tokens and $50 per million output, double Opus 4.8's $5 / $25 and five times Sonnet 5's introductory $2 / $10 (through August 31, 2026). Those rates are about to matter more: inside the Claude subscription Fable was capped at 50% of weekly usage limits, and on July 7, 2026 it leaves the subscription entirely. After that, using it means paying the API price.

You should still use Fable where it earns its price. But with the subscription cushion gone, every token lands on the bill, so a few precautions go a long way.

Where the money goes

Output tokens carry the higher unit price, $50 per million on Fable against $10 for input. But an agentic loop feeds in far more than it generates: the whole conversation is re-sent on every turn, so a ten-step task reprocesses the same context ten times. In our token benchmark, each documentation query ran 80,000 to 140,000 tokens, only one to three thousand of them output. Both dimensions cost money, so the levers below split in two.

To cut input tokens:

Start a fresh session for unrelated work, so old context stops being re-sent.

Keep prompt caching intact, so the re-sent prefix bills at a tenth of the price.

Trim the transcript so stale content stops riding along.

Hand the agent ready-made context instead of making it explore: skills for your own knowledge, Context7 for library docs.

To cut both input and output:

Match effort to the task; it governs thinking, output, and tool-call spend.

Tier your models, so only the work that needs Fable runs on Fable.

Tier your models: Fable to think, cheaper models to do

Most agentic work is a little hard reasoning wrapped in a lot of mechanical execution. Planning an approach or reviewing a diff for real bugs is where Fable earns its price; renaming symbols across forty files or running the test suite is not. Keep Fable as the orchestrator and hand the bulk work to Sonnet 5 or Haiku 4.5 subagents. Each runs in its own context window, so the cheap model never touches Fable's expensive context and Fable only sees the result. Delegate ten file edits and those loops run at $2 / $10 instead of $10 / $50.

In Claude Code, two features do this for you.

Subagents run on a model you choose: a small Markdown file under .claude/agents/ whose model: field takes a cheap alias like haiku or sonnet. Ask Claude to create one and it delegates automatically. The built-in Explore agent is the read-only version, inheriting your main model capped at Opus, so a Fable session already searches on Opus; define your own Explore with model: haiku to go cheaper.

Workflows set the model and effort per stage: run a review's adversarial find stage on a cheap model and reserve Fable for the synthesis that needs judgment.

The intelligence lives in the orchestration. Put Fable there, and push everything else to the cheapest model that can do the job.

Match effort to the task

Fable's effort parameter controls how much it thinks and how many tool calls it makes, and Anthropic treats it as the main lever for cost control on adaptive-thinking models. At max and xhigh, Fable can deliberate well past what a routine task needs.

So do not default to the ceiling. Anthropic notes that lower effort on Fable still performs well and often beats the xhigh output of prior models: reserve xhigh for genuinely hard work, keep high for intelligence-sensitive tasks, and drop to low or medium for routine steps. Dropping a level on a task that already works costs you nothing. Set it with /effort in Claude Code, or output_config on the API.

Keep prompt caching intact

Prompt caching turns repeated context into reads that cost about a tenth of the base input price, which matters most on Fable since its whole prefix is re-sent every turn.

Claude Code sets it up for you, so the job is mostly not breaking it. The cache keys off a stable prefix (tools, then system prompt, then conversation); change something early and everything after is re-read at full price. The easy way to trigger that is changing the tool set mid-task:

Adding or removing an MCP server changes the tools and invalidates the whole cache. Set servers up before you start, not partway...