Claude Sonnet vs Opus vs Haiku vs Fable: Which Model to Use

As of October 8, 2026, the current Claude lineup is Haiku 5.5, Sonnet 5.5, Opus 5.5 and Fable 5.1. Haiku 5.5 arrived on October 7 and Sonnet 5.5 on September 28, so Haiku 4.5 and Sonnet 5 are now legacy models that still work but are no longer the ones to start with. The four run from fastest and cheapest (Haiku) to most capable, slowest and most expensive (Fable). The model to start with depends on how you pay for Claude:

• Claude app on Free, Pro or Max: start with Sonnet. It uses a moderate share of your usage limit and handles coding, writing and analysis. Use Haiku for quick answers and summaries, Opus when Sonnet struggles, and Fable for long, accuracy-critical work (part of your limits on Max, usage credits on Pro).
• Claude Code: the account default is Opus 5.5 on Pro and Max, but Anthropic still recommends Sonnet for most coding because Opus uses several times more quota per turn. /model sonnet now selects Sonnet 5.5, and /model opusplan plans with Opus 5.5 and executes with Sonnet 5.5.
• Claude API: Anthropic's docs say most workloads start with Opus 5.5, or with Haiku 5.5 when cost and latency come first. Haiku 5.5 costs $0.10 / $0.50 per million input/output tokens for prompts up to 100,000 tokens and five times that above. Move to Fable 5.1 only when Opus 5.5 at xhigh or max effort still falls short on your tests.

On every surface, try a different effort level before you switch models. The sections below show the numbers behind each choice and the signals that tell you to move up or down.

Haiku 5.5, Sonnet 5.5, Opus 5.5 and Fable 5.1 at a glance

All figures are from Anthropic's models overview, API pricing page and plan pricing page as of October 8, 2026. API prices are standard first-party rates in US dollars per million tokens.

A few details in that table change real decisions:

• Haiku 5.5 is priced by prompt length. A prompt of up to 100,000 tokens pays $0.10 input, $0.01 cache read and $0.50 output. A longer prompt pays five times every rate: $0.50, $0.05 and $2.50. Anthropic's pricing page does not say whether cached tokens count toward that 100,000-token line, so test with your real prompts if you sit near it. Below the line, Haiku 5.5 is 90% cheaper than Haiku 4.5 ($1 / $5); above it, the saving is 50%.
• Sonnet 5.5 keeps Sonnet 5's $2 / $10 but halves cache reads to $0.10. A cache read is a part of the prompt that Claude already stored from an earlier request, such as a long system prompt or a codebase an agent keeps re-reading. Sonnet 5 charged $0.20 for that, the same as Opus 5.5. Now every Opus 5.5 rate is exactly double the Sonnet 5.5 rate, so cache-heavy work no longer narrows the gap between them (see the cost examples below).
• All four models now share a 1M-token context window and 128K max output. The old reason to rule out Haiku for long inputs is gone. Legacy Haiku 4.5 still has 200K context and 64K output.
• Default effort differs by model. Opus 5.5 and Haiku 5.5 run at medium unless you set it, while Sonnet 5.5 and Fable 5.1 run at high. Comparing models at their defaults therefore mixes a model difference with an effort difference.
• Moving from Haiku 4.5 changes your token counts. Haiku 5.5 uses the same newer tokenizer as the other three current models, so the same text counts as about 30% more tokens than it did on Haiku 4.5. Anthropic's Haiku 5.5 migration guide says to recount prompts with the new model rather than reuse old counts.

Opus 5.5 was released on September 22, 2026. Sonnet 5, Haiku 4.5, Opus 5, Fable 5 and older models are listed as legacy models and remain available. If you are moving from Opus 5, the Claude Opus 5.5 guide to API price, model ID and migration checks covers the switch. Haiku 4.5 has no deprecation notice as of October 8, 2026, and its retirement commitment is "not sooner than October 15, 2026." That is a floor, not a scheduled shutdown date. If you run Haiku 4.5 at volume, plan the move anyway: Haiku 5.5 thinks by default, and a non-default temperature, top_p or top_k returns a 400 error (what's new in Haiku 5.5).

Which Claude model to start with: Sonnet in the app, Opus 5.5 on the API

Anthropic gives two different starting points, and both are correct for their audience. A subscriber spends a shared usage limit, so the lighter model that still does the job is the better default. An API developer pays per token and can test quality against their own evaluation set, so starting strong and then cutting cost is a reasonable plan.

In the Claude app: Sonnet first

Anthropic's Claude Academy tutorial ranks the models by how much of your rate limit they use: Haiku is the lightest, Sonnet moderate, Opus heavy, and Fable the heaviest. It calls Sonnet "your versatile default" and says, "If you're not sure which model to pick, start here." Its own examples map everyday jobs to models:

The tutorial and the plan page name model families, not version numbers, so this ranking carries over to Haiku 5.5 and Sonnet 5.5.

Anthropic publishes only the order of limit use, not a multiplier. There is no official figure for how many Sonnet messages equal one Opus or Fable message. Usage depends on conversation length, the model, the features you use and the effort level.

In Claude Code: the default is Opus 5.5, the everyday choice is Sonnet 5.5

Since Claude Code v2.1.280, the default model setting resolves to Opus 5.5 on Pro, Max, Team, Enterprise and the Anthropic API, according to the Claude Code model configuration docs. Before that version, Pro and Team Standard users started on Sonnet 5, and Max users on Opus 5.

That default is not advice to keep Opus on for everything. Anthropic's Claude Code help page on models and limits says "Sonnet is the right choice for the large majority of coding work" and that "Opus costs several times more per turn than Sonnet, and Sonnet more than Haiku." It suggests Opus when you are genuinely stuck or the change is wide, and Haiku for renames, log lines and boilerplate.

The /model aliases on the Anthropic API:

Switching models mid-session keeps the conversation. Fable is never the account default, so you have to pick it. If it does not appear in /model, see Fable 5.1 Not Available in Claude Code? Diagnose It by Symptom.

On the API: Opus 5.5 for hard work, Haiku 5.5 for volume

Anthropic's model selection guide says, "Most workloads start with Claude Opus 5.5." It describes two ways to begin:

• Efficiency-first: build with Haiku 5.5 and upgrade only where you find a capability gap. This suits prototyping, tight latency budgets, cost-sensitive products and high-volume, straightforward tasks.
• Capability-first: build with Opus 5.5, then cut cost by lowering effort or moving parts of the work to a cheaper model. If your evals at xhigh or max effort still fall short on demanding reasoning or long-horizon agent work, move to Fable 5.1.

Sonnet 5.5 sits between them. Anthropic lists it for everyday coding, agent and enterprise workloads that need both speed and capability. Haiku 5.5 is its pick for the lowest latency and price: real-time features, high-volume pipelines and sub-agent tasks.

You also do not have to pick one model per product. A common mix is an executor model that escalates hard decisions to a stronger one, or an orchestrator that hands bulk work to cheaper workers such as Haiku 5.5 sub-agents. Claude Code's opusplan is a built-in example of the first pattern.

If you are choosing a cheap model across vendors rather than within Claude, Claude Haiku 5.5 vs GPT-6 Luna Pricing: Same Until 100K, Then 5x compares Haiku 5.5's two price tiers with OpenAI's cheapest model.

Which plan gives you which model: Opus from Pro, Fable on Max or credits

Source: claude.com/pricing for model access (as of October 8, 2026) and plan prices (as of September 24, 2026), plus Anthropic's help article on Fable models by plan. Prices exclude tax. Team and Enterprise plans list all four model families.

Every plan has a rolling five-hour session limit, and paid plans add a weekly limit. Claude on the web, desktop, mobile and Claude Code all draw from the same pool, so heavy Opus use in Claude Code leaves less for chat.

The Fable rules trip people up:

• Pro: Fable 5.1 is not part of your plan's limits. It runs on usage credits, which are billed at standard API rates, from the first request.
• Max: Fable 5.1 can use up to half of your weekly limit at no extra cost. That half comes out of your regular weekly limit and is used up faster than with other models. It is not extra capacity.

For the details, see Claude Fable 5.1 Usage Limits: Weekly Allowance, Extra Usage, and Model Switching and Claude Code Pricing: Pro vs Max and When the Upgrade Pays Off. If you are deciding which plan to buy, Claude Cowork Pricing: What You Actually Pay on Each Plan compares the plans.

One more behavior can look like a model switch you did not ask for. For a narrow set of biology and cybersecurity requests, Opus 5.5 and Fable conversations can fall back to an older Opus model. It is a safety measure and does not mean your plan lost access. Haiku 5.5 also runs safety classifiers, but it has no such fallback: a flagged request is declined instead.

When to move up a model, and when to move down: effort first

Effort is the setting that controls how much Claude thinks and how thorough it is. On the API it has five levels: low, medium, high, xhigh and max, and all four current models support every level (effort docs). The Claude app model picker has its own effort control. Higher effort usually helps on hard problems and uses more tokens, which means a bigger bill or more of your limit. Effort affects every output token, including thinking and tool calls.

Anthropic's guidance puts it plainly: "Tuning effort is often a better lever than switching models." Its starting points differ by model:

Two warnings from the same docs. At low, Haiku 5.5 may skip a search or stop early on long agent prompts. And Sonnet 5.5's effort levels are recalibrated compared with Sonnet 5, so a level you tuned on Sonnet 5 needs a fresh sweep. If Haiku 5.5 at xhigh or max still falls short, compare it with Sonnet 5.5 before raising effort further.

A practical sequence:

1. Define a check before you compare. Use a test suite that must pass, fields that must be extracted correctly, or a task you already know the answer to. Without a check, you are comparing impressions.
2. Move up when the current setup keeps failing that check. Raise effort on the same model first. On Opus 5.5 the default is medium, so high and xhigh are the obvious next steps. Switch to the next model only if the failures persist.
3. Move down when a cheaper model passes the same check. The Academy tutorial suggests running the same known task on Haiku and then Sonnet in new chats and comparing where the answers differ, not how long they are. On the API, run the same eval set at a lower effort or on the cheaper model.
4. Split the work if only part of it is hard. Plan with Opus and execute with Sonnet, or route simple subtasks to Haiku.

The Opus 5.5 effort guide shows how to set each level in Claude Code and the API and how to measure the cost.

Opus 5.5 or Fable 5.1 for the hardest work

This is the escalation question with the least obvious answer. Fable 5.1 costs 2.5 times as much as Opus 5.5 per input and output token. Anthropic's Opus 5.5 announcement claims that Opus 5.5 "performs at the level of Claude Fable 5.1 on most work." It adds that in Anthropic's own use, the gap between the two is narrower than benchmark scores suggest. In one internal test, both models translated HAProxy from C to Rust and passed nearly all of its regression tests. Opus 5.5 took 9.5 hours against 12 for Fable 5.1, at 51% less cost. These are Anthropic's claims and a single internal test, not an independent measurement.

Fable 5.1 still has a clear role in Anthropic's docs. It covers agent sessions that run for hours and research carried through to a finished document, spreadsheet or deck. It is also the next step when Opus 5.5 at xhigh or max still falls short. The way to find out is a matched trial on your own hard tasks:

Judge the trial by cost per accepted task, not cost per call:

cost per accepted task = (token costs + tool charges + retries + review and repair time) ÷ tasks that passed the check

Count every attempt, including timeouts and unusable answers. Fable 5.1 earns a place in your routing only if it passes checks that Opus 5.5 misses, or if it needs so few retries that it costs less per accepted result despite its higher rates. For Fable's full specs, see Claude Fable 5.1: API Pricing, Changes, and Migration Guide.

What the same work costs on the API: Haiku 5.5 in fractions of a cent

Per-token prices only tell you part of the story, because your token mix decides how far apart the models land. The two examples below use the standard rates from the table above and assume each model uses exactly the same number of tokens. They are illustrations, not measurements. Tokens are counted in millions, so 30,000 tokens is 0.03.

Example A: one request with a 90,000-token prompt. 30,000 fresh input tokens + 60,000 cache-read tokens + 5,000 output tokens, with no cache writes or tool charges. At 90,000 tokens the prompt stays inside Haiku 5.5's lower price tier:

• Haiku 5.5: 0.03 × $0.10 + 0.06 × $0.01 + 0.005 × $0.50 = $0.0061
• Sonnet 5.5: 0.03 × $2 + 0.06 × $0.10 + 0.005 × $10 = $0.116
• Opus 5.5: 0.03 × $4 + 0.06 × $0.20 + 0.005 × $20 = $0.232
• Fable 5.1: 0.03 × $10 + 0.06 × $0.25 + 0.005 × $50 = $0.565

Example B: a cache-heavy agent run. 20 turns that each re-read the same 80,000-token cached prefix (1,600,000 cache-read tokens in total, with every request kept under 100,000 tokens), plus 50,000 new input tokens and 10,000 output tokens:

• Haiku 5.5: 1.6 × $0.01 + 0.05 × $0.10 + 0.01 × $0.50 = $0.026
• Sonnet 5.5: 1.6 × $0.10 + 0.05 × $2 + 0.01 × $10 = $0.36
• Opus 5.5: 1.6 × $0.20 + 0.05 × $4 + 0.01 × $20 = $0.72
• Fable 5.1: 1.6 × $0.25 + 0.05 × $10 + 0.01 × $50 = $1.40

What the two examples show:

• Sonnet 5.5 costs about 19 times as much as Haiku 5.5 in Example A and about 14 times in Example B. As long as prompts stay under 100,000 tokens, Haiku 5.5 is in a different price class from the rest of the lineup.
• Opus 5.5 costs exactly twice as much as Sonnet 5.5 in both examples. Every Opus 5.5 rate is double the Sonnet 5.5 rate, including cache reads, so no token mix changes that ratio at equal token counts.
• Fable 5.1 costs about 2.4 times as much as Opus 5.5 in Example A and about 1.9 times in Example B. At equal token counts it is the most expensive model in every category.
• The legacy models cost more for the same tokens. Haiku 4.5 comes to $0.061 in Example A and $0.26 in Example B, ten times Haiku 5.5. Sonnet 5 comes to $0.122 and $0.52; the whole difference from Sonnet 5.5 is the halved cache-read rate.

Haiku 5.5's advantage shrinks once a prompt passes 100,000 tokens. Run Example A with a 150,000-token prompt (90,000 fresh + 60,000 cache reads + 5,000 output): Haiku 5.5 moves to its higher tier and costs 0.09 × $0.50 + 0.06 × $0.05 + 0.005 × $2.50 = $0.0605, against $0.236 on Sonnet 5.5. It is still about a quarter of Sonnet's cost, but ten times its own price below the line. That calculation assumes cached tokens count toward the 100,000-token threshold, which Anthropic does not state either way.

Example B leaves out the first cache write. Writing 80,000 tokens to the five-minute cache adds 0.08 × the write rate: $0.01 on Haiku 5.5, $0.20 on Sonnet 5.5, $0.40 on Opus 5.5 and $1.00 on Fable 5.1.

Real tasks do not use equal tokens. Higher effort produces more thinking tokens. A stronger model may finish in fewer turns, and output lengths differ from model to model. A cheaper model that needs three retries can end up costing more per finished task. The examples tell you where the rate card puts each model. Only your own workload tells you the cost of a finished task.

Other pricing options change the numbers but not the order. The Batch API halves input and output prices: $0.05 / $0.25 for Haiku 5.5 up to 100,000-token prompts ($0.25 / $1.25 above), $1 / $5 for Sonnet 5.5, $2 / $10 for Opus 5.5 and $5 / $25 for Fable 5.1. Opus 5.5 also has a Fast mode research preview at $8 / $40, which should not be mixed into a standard-rate comparison. Sonnet 5.5, Opus 5.5 and Fable 5.1 bill their full 1M-token context at standard rates, with no long-context surcharge; Haiku 5.5 is the exception with its 100,000-token tier.

Quick answers about Haiku, Sonnet, Opus and Fable

What is the difference between Claude Sonnet, Opus and Haiku?

They are three sizes of the same model family, trading speed and price for capability. Haiku 5.5 is the fastest and cheapest, built for high-volume work such as classification, extraction and routing. Sonnet 5.5 balances speed and intelligence and is Anthropic's everyday default for coding, writing and analysis. Opus 5.5 is slower and costs twice as much per token as Sonnet 5.5, and it targets long-running agentic coding and knowledge work. All three have a 1M-token context window. Fable 5.1 sits above Opus as the most capable model.

Is Opus or Sonnet better for Claude Code and everyday work?

Opus 5.5 is the more capable model, but Sonnet is the better default for most work. Anthropic's Claude Code help page says Sonnet is right for the large majority of coding, and Opus uses several times more of your quota per turn. On the API, Opus 5.5 costs exactly twice Sonnet 5.5 at every rate. Switch to Opus when Sonnet keeps failing a task even at higher effort, or plan with Opus and execute with Sonnet through opusplan.

Is Claude Fable better than Sonnet?

Fable 5.1 is more capable than Sonnet 5.5, but it is slower and costs five times as much per API token ($10 / $50 against $2 / $10). It is not on the Free plan, runs on usage credits on Pro, and can use up to half of your weekly limit on Max. For everyday coding, writing and analysis, Sonnet is the better value. Reserve Fable for long, demanding reasoning and agent work that Opus 5.5 cannot finish.

Which Claude model is the best?

Fable 5.1 is Anthropic's highest-capability model, but it is rarely the best first choice. Anthropic says Opus 5.5 performs at Fable's level on most work at a lower price. For everyday tasks, Sonnet 5.5 usually gives the best balance of quality, speed and limit use. The best model is the cheapest one that passes your check.

When should you use Fable instead of Opus?

Use Fable 5.1 when Opus 5.5 at a high effort level (xhigh or max) still fails a task that matters. Typical cases are agent sessions that run for hours, or research that has to end in a finished document. Confirm the choice with a matched trial on your own tasks rather than benchmark scores.

Is Haiku faster than Sonnet?

Yes. Anthropic lists Haiku 5.5 as the fastest model in the lineup and Sonnet 5.5 as fast, but it does not publish a speed multiplier. Haiku 5.5 thinks by default at medium effort, so for the lowest latency set effort to low for chat and short tool tasks. On price, Haiku 5.5 costs $0.10 / $0.50 per million tokens up to 100,000-token prompts against $2 / $10 for Sonnet 5.5.

Which is better for writing, Haiku or Sonnet?

Sonnet. Anthropic's Academy tutorial maps writing, debugging and multi-step analysis to Sonnet, and keeps Haiku for summaries, quick answers and simple extraction. If you write in volume through the API, run a few known pieces through both and compare where the drafts differ before moving the bulk to Haiku 5.5.

What is the new Haiku model, and what is its API name?

The new model is Claude Haiku 5.5, released on October 7, 2026, with the API model ID claude-haiku-5-5. The previous model, Haiku 4.5, still works as a legacy model under claude-haiku-4-5-20251001 (alias claude-haiku-4-5). In Claude Code, the haiku alias now points to Haiku 5.5.