Claude Opus 5.5 vs Fable 5.1 vs Sonnet 5.5

In short

Start with Opus 5.5. Anthropic tells developers who are unsure to begin there, Claude Code uses it by default, and it costs $4 per million input tokens and $20 per million output tokens. Move fast, repetitive or high-volume work down to Sonnet 5.5, which responds faster and costs half as much ($2 and $10). Try Fable 5.1 only when Opus 5.5 at a higher effort setting still fails your own tests: it costs 2.5 times as much ($10 and $50), responds more slowly, runs on usage credits on Claude Pro and requires 30-day data retention on the API. All three have a 1M-token context window and 128K tokens of output, so size does not decide.

On this page4

Anthropic's own benchmarks put Opus 5.5 ahead of Fable 5.1

Anthropic says Opus 5.5 "performs at the level of Claude Fable 5.1 on most work", and its launch table scores Opus 5.5 above Fable 5.1 on every benchmark listed. Sonnet 5.5's launch table compares it with Opus 5.5 on the same benchmarks. Opus 5.5's scores are identical in both tables, so the three columns below can be read side by side.

Benchmark (Anthropic's figures)	Sonnet 5.5	Opus 5.5	Fable 5.1
Terminal-Bench 4.0	70.6	66.4	55.8
FrontierCode 1.1 Main	52.1	54.4	50.3
CursorBench 4.0	55.5	57.8	51.8
GDPval-AA v2.1 (Elo)	1844	1846	1735
HLE with tools	64.5	67.7	65.6
OSWorld 2.1 partial	80.1	81.8	80.7

Sonnet 5.5 beats Opus 5.5 only on Terminal-Bench 4.0. On FrontierCode, CursorBench, HLE and OSWorld, Opus 5.5 is ahead by 1.7 to 3.2 points, and the two are level on GDPval-AA. Fable 5.1 trails Opus 5.5 on every row. Anthropic ran Opus 5.5 at max effort except on Terminal-Bench 4.0 (xhigh), and Sonnet 5.5's FrontierCode score is at xhigh (46.2 at max).

These are the vendor's own tests. Anthropic still describes Fable 5.1 as the model "for demanding reasoning and long-horizon agentic work" and suggests it when "your evals on Claude Opus 5.5 at higher effort still fall short". Fable 5.1 is something to test against a real failure, not a default upgrade.

Sonnet 5.5 costs half as much as Opus 5.5; Fable 5.1 costs 2.5 times as much

Prices are per million tokens on Anthropic's API. Read across a row: each step up at least doubles the rate, except for cache reads, where Fable 5.1 costs only a quarter more than Opus 5.5.

	Sonnet 5.5	Opus 5.5	Fable 5.1
Input	$2	$4	$10
Output	$10	$20	$50
Cache read	$0.10	$0.20	$0.25
Batch input / output	$1 / $5	$2 / $10	$5 / $25
Anthropic's latency label	Fast	Moderate	Slower

List prices are not costs per task. Thinking tokens are billed as output, and the default effort differs: on the API, Opus 5.5 defaults to medium while Sonnet 5.5 and Fable 5.1 default to high. A Sonnet 5.5 call at high can therefore write more output than the same Opus 5.5 call at medium, and Anthropic calls effort "a behavioral signal, not a strict token budget". Set effort explicitly and compare the cost of finished tasks.

Fast mode exists only on Opus. It gives up to 2.5 times the output speed at $8 input and $40 output for Opus 5.5, as a research preview on Anthropic's own API that needs access approval. In Claude Code on a subscription it runs on usage credits only.

Fable 5.1 is outside the plan on Pro and never the Claude Code default

Anthropic's pricing page lists model families by plan. The table shows which plans include each family, and where Fable runs on usage credits instead of the plan's limits.

	Sonnet	Opus	Fable
Free	Yes	No	No
Pro ($20 a month)	Yes	Yes	Usage credits only
Max (from $100 a month)	Yes	Yes	Up to 50% of weekly limits
Team	Yes	Yes	Premium seats; usage credits on standard seats
Claude Code	`/model sonnet`	Default	`/model fable`

On Max and Team premium seats, Fable use counts against the normal weekly limit, up to half of it; Anthropic stresses this is "not a 50% increase". Claude Code uses Opus 5.5 by default on Pro, Max, Team, Enterprise and the API, and "Fable models are not the default on any plan or provider." Where Fable bills usage credits, the model picker says "Requires usage credits". Requests that Fable's safety classifiers flag are re-run on another model (biology on Opus 5, cybersecurity on Opus 4.8). On Amazon Bedrock and Google Cloud, Claude Code's sonnet alias still points to Sonnet 4.5, so set the full model ID there. Those clouds also set their own prices.

Data handling differs too. On the API, Fable 5.1 requires 30-day retention, and zero data retention is not available for it unless Anthropic expressly authorizes it. Opus 5.5 and Sonnet 5.5 prompts and outputs are not retained by default.

Checks before moving work off Opus 5.5

Up to Fable 5.1: first re-run the failing tasks on Opus 5.5 at high, xhigh or max. Then price the move per finished task, remembering that Fable's output costs $50 per million and thinking counts as output. Confirm that 30-day retention is acceptable, and on Pro or a Team standard seat, that you will pay usage credits.
Down to Sonnet 5.5: set effort explicitly. If you want it to act without thinking first, between_tools turns off up-front thinking at high effort or below.
Further down: for classification, extraction and routing Anthropic names Haiku 5.5 instead; see Claude Haiku 5.5 vs GPT-6 Luna.

Claude Sonnet 5.5 vs Opus 5.5 covers the lower step in more detail, and Claude usage limits by plan explains the five-hour and weekly limits these plans share.

Official sources

Editorial guidance grounded in official product sources.

Show all 24 sources

FAQ

Common questions

Which Claude 5 model should most people start with?

Start with Opus 5.5 ($4/$20): Anthropic tells developers who are unsure to begin there, and Claude Code uses it by default. Move latency-sensitive, repeatable or high-volume work down to Sonnet 5.5, which is faster and costs half as much ($2/$10). Fable 5.1 is an escalation for when Opus 5.5 at higher effort still falls short, and it must prove extra value on a matched evaluation.

Is Fable 5.1 worth 2.5 times Opus 5.5's API price?

Only a representative workload can establish value. Fable 5.1 has 2.5 times Opus 5.5's standard input/output rates ($10/$50 vs $4/$20), and its cache reads cost US$0.25 versus Opus's US$0.20 per million tokens. Compare total task cost, successful completion and review effort on real tasks.

Do Fable 5.1, Opus 5.5, and Sonnet 5.5 have different context limits?

No. Anthropic lists a 1-million-token context window and a 128,000-token maximum output for all three current models. Context size therefore does not decide among them; capability, latency, effort, price, access, retention, and safeguard behavior do.

Can I use Sonnet 5.5, Opus 5.5, and Fable 5.1 in Claude Code?

Yes, subject to the account and provider. Opus 5.5 is the Claude Code default. Sonnet 5.5 requires Claude Code 2.1.284 or later and is what the sonnet alias selects on the Anthropic API; other providers map that alias to older Sonnet versions. Fable 5.1 needs Claude Code 2.1.257 or later and is selected through the fable alias or full model ID; it is not the default on any plan. Check the actual model, subscription allowance and usage-credit/API route.

Can eligible ZDR customers use Fable 5.1?

Usually not. Fable 5.1 requires 30-day retention, so ZDR is not available for it unless Anthropic expressly authorizes it. Confirm that authorization in the agreement rather than assuming every ZDR account qualifies.

Should Opus 4.8 or Mythos 5.1 still be part of this choice?

Usually not. Opus 4.8 is a legacy model that gets no further updates, and at $5/$25 it costs more than Opus 5.5 at $4/$20. Mythos 5.1 has Fable 5.1's capabilities and price ($10/$50), but Anthropic offers it only to organizations verified through its verification programs, and it also requires 30-day retention. Neither should displace the current Sonnet 5.5, Opus 5.5 and Fable 5.1 selection.

Next steps

Claude Opus 5.5 vs Fable 5.1 vs Sonnet 5.5: Which Should You Use?