Pricing · verified 2 September 2026

Claude Fable 5.1 pricing

Cache reads fall to $0.25 per 1M tokens, a 75% reduction from Claude Fable 5. List price is unchanged at $10 input and $50 output. The cache arithmetic reproduces Anthropic's "25% typical / 45% agentic" claim from the published rates alone.

All six billing dimensions

Claude Fable 5.1 prices input, output, cache read, cache write (five-minute and one-hour tiers), and batch. List prices match Anthropic, AWS Bedrock, Microsoft Foundry and OpenRouter; Google Vertex carries an 11% premium reflecting the data-residency multiplier.

DimensionClaude Fable 5.1Claude Fable 5Notes
Input / 1M tokens$10.00$10.00Identical across providers (Google Vertex +10%)
Output / 1M tokens$50.00$50.00Identical across providers (Google Vertex +10%)
Cache read / 1M tokens$0.25$1.0075% reduction; the entire commercial case
Cache write 5-min / 1M$12.50$12.50Break-even is a single re-read
Cache write 1-hour / 1M$20.00$20.00For long-running agents
Batch input / output per 1M$5.00 / $25.00$5.00 / $25.00Halves both directions

The cache arithmetic, reproduced from the published rates

Cost equals uncached input times the input rate plus cached input times the cache-read rate plus output times the output rate. Cache WRITE cost is excluded because a long-running agent writes the prefix once and reads it many times; on a session with more than two reads the write is immaterial. State this assumption on the page.

Workload profileInput tokensCache hit rateOutput tokensFable 5.1 cost / taskFable 5 cost / taskSaving %
A. Chat / short prompt, little reuse200,00020%20,000$2.61$2.641.1%
B. Typical agentic session10,000,00090%200,000$22.25$29.0023.3%
C. Long-running cache-heavy agent20,000,00095%100,000$19.75$34.0041.9%

What this reproduces

Profiles B and C reproduce Anthropic's "about 25% cheaper for typical workloads, up to about 45% for agentic work" claim from the published rates alone, which means OneInfer can state it as a verified calculation with its own method rather than as a vendor quote. Profile A is the honest counterweight: on short, low-reuse prompts the 75% cache cut is nearly invisible. Publishing profile A is what makes profiles B and C credible.

Ready to test the workflow?

Create account & add credits

Cost per benchmark point

On cost per Terminal-Bench 4.0 point, Claude Opus 5 beats Claude Fable 5.1. Fable 5.1 buys the top score and the science capability, not the best value per point. A comparison page that concludes everything favours the newest model is not a comparison page and will not be cited.

ModelInput $/1MOutput $/1MTask costTerminal-Bench 4.0Cost per TB4.0 point
Claude Fable 5.1$10$50$2.0055.8%$0.0358
Claude Fable 5$10$50$2.0042.0%$0.0476
Claude Opus 5$5$25$1.0052.3%$0.0191
Claude Mythos 5.1$10$50$2.0060.9%$0.0328
GPT-5.6 Sol$5$30$1.1037.3%$0.0295

Frequently asked questions

How much does Claude Fable 5.1 cost?

Claude Fable 5.1 costs $10 per 1M input tokens and $50 per 1M output tokens on OneInfer, matching Anthropic's list price. Cached input reads at $0.25 per 1M tokens, a 75% reduction from the $1.00 that Claude Fable 5 charges. Cache writes cost $12.50 per 1M for the five-minute tier and $20.00 for the one-hour tier. Batch requests halve both directions to $5.00 and $25.00.

Which is cheaper, Claude Fable 5.1 or Claude Fable 5?

It depends entirely on how much cached context you re-read per token generated, because list prices are identical at $10 per 1M input and $50 per 1M output. The only difference is the cache read rate, $0.25 against $1.00. On a short prompt at a 20% hit rate with little reuse, the saving is about 1% and not worth a migration. On a long-running agent at a 95% hit rate, total task cost falls from $34.00 to $19.75, a saving of about 42%.

Put Claude Fable 5.1 to work

Fund a controlled evaluation, start with a prepared prompt, and measure quality and cost on your own workload.