Claude Sonnet 5.5 API: Pricing and Quickstart

Q: What is the model ID?

claude-sonnet-5-5 (hyphens, dots can break in some clients).

Claude Sonnet 5.5 is Anthropic's current daily-driver model, released on September 28, 2026. It replaces the previous Sonnet 4.6 with a 33% price drop and meaningful capability gains in coding and reasoning.

Google's autocomplete for "claude sonnet 5.5" already surfaces "api pricing" and "cost" as top suggestions, which tells you what people are looking for.

Pricing

Item	Per million tokens
Input	$2
Output	$10
Cache write	$2
Cached read	$0.20

Compared to the previous Sonnet 4.6 at $3/$15, input and output both dropped 33%. Compared to the heavier Opus 5.5 at $4/$20, Sonnet covers most daily work at half the price.

At $2/$10 it goes head-to-head with GPT-6.1 Sol, which shares the exact same official pricing. The most direct cross-provider comparison in the current lineup.

Specs

Context: 1,000,000 tokens (1 million)
Max output: 128,000 tokens
Protocols: Anthropic Messages (/v1/messages) and OpenAI Chat Completions
Capabilities: tool calling, structured output, multimodal input
Model ID: claude-sonnet-5-5

Connecting

Option 1: Anthropic Messages API (native)

curl -s https://api.hiapi.ai/v1/messages \
  -H "x-api-key: YOUR_HIAPI_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5-5",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Explain the difference between Sonnet 5.5 and Opus 5.5 in three sentences"}]
  }'

Option 2: OpenAI Chat Completions (compatible)

curl -s https://api.hiapi.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_HIAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5-5",
    "messages": [{"role": "user", "content": "Explain the difference between Sonnet 5.5 and Opus 5.5 in three sentences"}]
  }'

Both use the same key. Messages is the native protocol; Chat Completions shares parsing code across Claude, GPT, and DeepSeek models.

Option 3: Claude Code

Two environment variables point Claude Code at a HiAPI key:

export ANTHROPIC_BASE_URL=https://api.hiapi.ai
export ANTHROPIC_AUTH_TOKEN=YOUR_HIAPI_API_KEY

After that reasoning inside Claude Code routes through HiAPI. See the Claude Code and Codex setup guide for the full walkthrough.

Sonnet 5.5 versus Sonnet 4.6

	Sonnet 5.5	Sonnet 4.6
Input	$2	$3
Output	$10	$15
Cached read	$0.20	$0.30
Context	1M	—
Coding/reasoning	Stronger	—

If you are on the old Sonnet 4.6, the upgrade is a single model-ID change with no parsing rewrite. The capability jump pays for the switch hundreds of times over.

When to use Sonnet 5.5 versus Opus 5.5

Use Sonnet 5.5: daily coding, document analysis, general Q&A. Output quality holds up here and it costs half of Opus.

Use Opus 5.5: long-running agent tasks, multi-step reasoning that Sonnet fumbles, low-tolerance work.

A productive habit: run a pass on Sonnet 5.5 first. Only escalate to Opus 5.5 when Sonnet genuinely cannot handle it. Most of the time you will not need to.

FAQ

Is Sonnet 5.5 available on HiAPI?

As of this writing, Sonnet 5.5 is not yet on HiAPI. Check the model catalog; it will appear when available.

What is the model ID?

claude-sonnet-5-5 (hyphens, dots can break in some clients).

Sonnet 5.5 versus GPT-6.1 Sol, which wins?

Same official price at $2/$10. No universal answer. Run your own typical tasks through both and compare output quality and token consumption.

What should I cache?

System prompts, long documents, conversation history. Cache writes are a one-time cost; subsequent reads cost $0.20. It pays back fast.

Do Sonnet 5.5 and Opus 5.5 share a context window?

No. Switching models starts a new session.

Bottom line

Sonnet 5.5 is the equilibrium point in Claude's lineup: stronger than the generation it replaces, half the price of the flagship, sufficient for most work. Model ID is claude-sonnet-5-5. Both protocols connect in the same way every Claude model does.

Sources