Opus 5.5 and Sonnet 5.5 are priced on the same structure: same 1M context, same 5%-of-input cache hits, same standard-rate long context. Opus simply costs twice as much on every line.
Claude Sonnet 5.5 is exactly half the price of Opus 5.5 on input, output, cache writes and cache hits. Because the ratio is constant, the decision is purely about quality: use Opus 5.5 where it saves retries or review time worth more than double the tokens.
| Claude Opus 5.5 | Claude Sonnet 5.5 | |
|---|---|---|
| Provider | Anthropic | Anthropic |
| Tier | Flagship | Mid-tier |
| Input / 1M tokens | $4.00 | $2.00 |
| Output / 1M tokens | $20.00 | $10.00 |
| Cache hit / 1M | $0.20 | $0.10 |
| Cache write / 1M | $5.00 | $2.50 |
| Context window | 1M | 1M |
| Notes | New leading Opus model (Sep 22, 2026); Fable 5.1-level performance at 20% less than Opus 5; included on Pro/Max/Team at no extra cost | Released Sep 28, 2026, replacing Sonnet 5 at the identical $2/$10 per 1M price; Sonnet 5 is now a legacy model |
Claude Sonnet 5.5 is 2.0× cheaper on input tokens ($2.00 vs $4.00 per 1M). Claude Sonnet 5.5 is 2.0× cheaper on output tokens ($10.00 vs $20.00 per 1M).
| Workload | Claude Opus 5.5 | Claude Sonnet 5.5 | Difference |
|---|---|---|---|
| Support chatbot 10,000 conversations × (2,000 input + 400 output tokens) | $160 | $80.00 | Claude Sonnet 5.5 saves $80.00 |
| Document Q&A 1,000 questions × (30,000 input + 800 output tokens) | $136 | $68.00 | Claude Sonnet 5.5 saves $68.00 |
| Coding agent session 4M input tokens (90% cache hits) + 150,000 output tokens | $5.32 | $2.66 | Claude Sonnet 5.5 saves $2.66 |
Standard rates per 1M tokens from data on this site. The coding-agent row assumes 90% of input is read from the prompt cache (full input price where a model has no cache rate). Cache-write fees, batch discounts and long-context surcharges listed in the notes are not included. Updated November 2, 2026.
the task is hard enough that Sonnet 5.5 needs retries, longer prompts or human fixes — complex multi-step agents, difficult refactors — and one Opus pass is cheaper than several Sonnet passes.
you want the default workhorse: it handles most production traffic at half the cost, and it replaced Sonnet 5 at the same price.
Because every Opus 5.5 rate is double Sonnet 5.5's — $0.20 vs $0.10 on cache hits, $5.00 vs $2.50 on writes — caching, batch and long prompts don't change the ratio. Opus is worth it whenever it finishes a task in fewer than half the tokens Sonnet needs, counting retries.
A common pattern is to run Sonnet 5.5 by default and escalate to Opus 5.5 when a check fails — failing tests, a validator rejecting JSON, low self-reported confidence. If 15% of requests escalate, the blended cost is about 1.15–1.3× Sonnet rather than 2×, while the hard cases still get Opus.
Anthropic's pricing page says both bill the full 1M-token context at standard rates and both get 50% off through the Batch API. Sonnet 5 is now a legacy model, so new Sonnet work should target 5.5. On the consumer side, Opus 5.5 is included on Claude Pro, Max and Team at no extra cost.
Exactly 2× on every rate: $4.00/$20.00 vs $2.00/$10.00 per 1M tokens.
No. Sonnet 5.5 replaced Sonnet 5 at the identical per-token input and output price, with a lower cache-hit rate.
No. Both bill the full 1M-token context at standard rates.
How we choose: prices come from each vendor's official pricing page and are re-checked every week. We take no affiliate fees, so rankings aren't for sale.