Claude Sonnet 5.5 API: $2/$10 Pricing and Setup

Claude Sonnet 5.5 is Anthropic's second Claude 5.5 model, released 28 September 2026. The Claude API id is claude-sonnet-5-5. Claude Sonnet 5.5 API pricing is $2 per 1M input tokens and $10 per 1M output tokens, the same sticker as Sonnet 5, with cache reads at $0.20 per 1M.

Sonnet 5.5 is not on RouterPlex yet. Create a RouterPlex account to use the current 60+ catalog models today, including Claude Sonnet 5 at those same $2/$10 rates. The new id works here once the marketplace lists it. One prepaid key, 0% markup on vendor list price, a $0 top-up fee, and a hard spend limit that stops at $0.

Claude Sonnet 5.5 API pricing #

USD per 1M tokens on Anthropic's platform. The announcement table's cache-write cell of $2.50 is the 5-minute rate. The overview also publishes a 1-hour cache write.

Anthropic's Batch API is 50% off input and output on that platform. Prompt caching up to 90% is also an Anthropic platform claim. Neither discount is a RouterPlex bill, because this model is not listed here.

A 40,000-token input and 2,000-token output, with no cache, is $0.08 + $0.02 = $0.10 on this sheet. That is the same arithmetic as live claude-sonnet-5 on RouterPlex. The 5.5 id is a different model.

The minimum cacheable prompt is 512 tokens.

What Sonnet 5.5 is #

Anthropic positions it for well-scoped everyday work next to Opus 5.5, which stays the stronger model for open-ended jobs. Zero data retention is available, same as Opus 5.5 and Sonnet 5, on Anthropic's terms.

If you currently disable thinking, switch to between_tools before moving. It works at high effort or below. Non-default temperature, top_p, or top_k returns HTTP 400.

Vendor benchmarks, 28 September 2026 #

These scores are Anthropic's. FrontierCode 1.1 Main is reported twice for Sonnet 5.5: 46.2% at Max and 52.1% at Xhigh. The other columns on the Xhigh row are blank on the announcement. Publish both effort labels. Do not treat 46.2% as the only FrontierCode score.

Terminal-Bench and OpenAI did not publish a GPT-6 Sol number, so Anthropic's chart uses GPT-5.6 Sol there. FrontierCode penalizes out-of-scope changes. Anthropic's footnote says that at Max, Sonnet 5.5 more often ran Claude Code's code-review skill, and in two cases Cognition examined that produced a timeout or extra edits, which lowered the score. Artificial Analysis ran GDPval-AA and AA-Briefcase on a pre-release deployment with a structured-output bug. Anthropic says to expect a small effect.

Other vendor claims from the same page, not RouterPlex measurements: at High effort on FrontierCode, about 10 points above Sonnet 5 at the same setting and about one fifteenth of the cost per task. At Medium effort, the apps default, Terminal-Bench far exceeds Sonnet 5's best for less than a tenth of the cost per task. Output is more than 30% faster, and cost per task is up to 30% lower because the model uses fewer tokens.

Daniel Vogel, COO at Epic Games, said early testing cleared the quality bar expected from a higher-tier model on a system design audit and a data flow review, including tens of thousands of lines and multi-hour tasks. Curtis Allen, Principal Engineer at Slack, said Slackbot offline evals improved with about 14% fewer output tokens and fewer steps, without prompt changes. Anthropic also says this is the first Sonnet to beat Pokemon Red from screenshots only.

Safeguards #

This is the first Sonnet with cyber safeguards and reasoning-extraction classifiers. Higher-risk cyber tasks fall back to Sonnet 5. Biology safeguards match Sonnet 5. Routine software work is the intended path. Details are in the system card.

Call Claude Sonnet 5 today #

claude-sonnet-5 is the live RouterPlex route at $2 / $10 per 1M. It is the previous Sonnet, not 5.5.

curl https://api.routerplex.com/v1/chat/completions \
  -H "Authorization: Bearer $ROUTERPLEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "messages": [
      {"role": "user", "content": "Name the riskiest assumption in this migration plan."}
    ]
  }'

The same key reaches the rest of the catalog through the OpenAI-compatible API. Compare gateways on the OpenRouter comparison. Opus 5 is live at $5 / $25. Fable 5 is live at $10 / $50.

On RouterPlex, once Sonnet 5.5 is listed #

When the row exists, the contract is vendor list price, 0% markup, prepaid balance, and a hard per-key budget. The request stops at $0. This page will name the RouterPlex route that day. Until then, paste claude-sonnet-5, not claude-sonnet-5-5.

Create a RouterPlex account if you want the key ready.