Claude offers three models with different strengths — Opus 5.5, Sonnet 5.5, and Haiku 5.5 (as of November 2026, all three tiers are on the 5.5 generation). The smartest model is not always the right answer. Pick based on the task's difficulty, speed, and cost. This guide lays out how the three differ and when to use which.
🟢 Model references match the current lineup · model notice · Fable subscription
Fable subscription (updated November 17, 2026): Claude Fable 5.2, released November 9, 2026, is the current Fable model; Fable 5.1 and Fable 5 are now legacy. Plan terms are the same for all three — Max and Ultra plans and premium seats on Team and seat-based Enterprise plans include Fable at up to 50% of the weekly usage limit; Pro plans and standard seats use usage credits (not available on Free). API pricing is $8/M input and $40/M output tokens for Fable 5.2, and $10/$50 for Fable 5.1 and 5. Some cybersecurity and biology requests may be answered by an Opus model due to safety classifiers. See the Fable 5.2 launch news and the Fable 5.2 guide for details.
The differences at a glance
Here are the core specs (Claude API, per the official Models and Pricing docs). Pricing is input / output per 1M tokens — a token being the unit Claude uses to process text; roughly one short word in English.
- Opus 5.5 (released September 22, 2026) — most capable. 1M context / 128K output tokens. $4 / $20.
- Sonnet 5.5 (released September 28, 2026) — balanced. 1M / 128K tokens. $2 / $10 (same as Sonnet 5; only cache reads dropped from $0.20 to $0.10).
- Haiku 5.5 (released October 7, 2026) — fastest. 1M / 128K tokens. Tiered pricing: $0.10 / $0.50 for prompts up to 100K tokens, $0.50 / $2.50 above that.
The previous generation — Opus 5 ($5 / $25), Sonnet 5 ($2 / $10), and Haiku 4.5 ($1 / $5, 200K context) — remains available but is classified as legacy in the official docs.
Note: pricing and specs follow the official Models and Pricing docs as of November 2026 and may change. Check the official docs for the latest.
Opus 5.5 — for hard work
Best for the most complex reasoning, long-horizon agentic work — autonomous, multi-step task execution using tools like browsers or terminals — coding, and tasks needing high autonomy. Anthropic says Opus 5.5 performs at the level of the higher-tier Fable 5.1 on most work and costs 40% less than Opus 5 on typical workloads. It is still twice the price of Sonnet 5.5 on both input and output, so it is overkill for light work.
Sonnet 5.5 — for most work
The best balance of capability and cost. It is priced the same as Sonnet 5, with cache reads at half the price. Most writing, coding, and analysis work is well served here — if you are unsure which model to use, start here.
Heads up: the Sonnet 5 family uses a new tokenizer — the engine that converts text into tokens — which produces roughly 30% more tokens for the same input than Sonnet 4.6 (Sonnet 5.5 shares Sonnet 5's tokenizer). If your cost estimates or token budgets are based on Sonnet 4.6, recount them. Also, on the API Sonnet 5.5 rejects thinking: {"type": "disabled"}, so code that turned thinking off on Sonnet 5 needs a change before you switch.
Haiku 5.5 — fast and lightweight
The fastest of the three, and the cheapest as long as prompts stay at or under 100K tokens. The official docs position it for high-volume, latency-sensitive work such as classification, extraction, and routing. Its context grew from Haiku 4.5's 200K to 1M tokens, but it uses a new tokenizer that counts roughly 30% more tokens than Haiku 4.5 for the same text — factor that into cost estimates.
How to choose
- Hard task, accuracy matters → Opus 5.5
- Everyday task, solid balance → Sonnet 5.5
- Speed, volume, cost savings → Haiku 5.5
That order is the cost-conscious approach. Anthropic's official model-selection guidance puts quality first: start most work on Opus 5.5 and use Fable 5.2 for demanding reasoning and long agentic tasks. Follow that if quality is your priority.
In claude.ai chat you can switch models in the interface; on the API — where your program calls Claude directly — you specify a model ID (claude-opus-5-5, claude-sonnet-5-5, claude-haiku-5-5). Above these three sits the higher tier Fable 5.2 (released November 9, 2026, $8 / $40), whose availability by plan differs from the standard models, so check its terms separately.