Claude Haiku 5.5 API: Model ID, Pricing, Claude Code Setup

Short answer

Claude Haiku 5.5 is available on TeamoRouter as claude-haiku-5-5, with a 1,000,000-token context window and up to 128,000 output tokens. Anthropic's list price is $0.10 per million input tokens and $0.50 per million output tokens — the cheapest current Claude model; on TeamoRouter it is about $0.03 / $0.15 as of the publish date. It is built for high-volume, latency-sensitive work: classification, extraction, routing, summaries and subagents. In Claude Code it belongs in the background slot: set ANTHROPIC_DEFAULT_HAIKU_MODEL=claude-haiku-5-5 and update Claude Code to 2.1.293 or newer.

What Haiku 5.5 is and when to use it

Haiku is Anthropic's smallest and fastest line. Anthropic released Haiku 5.5 on October 7, 2026, as the third model of the 5.5 family after Opus 5.5 and Sonnet 5.5, and describes it as the fastest model in the current lineup. It is also the first Haiku that takes the effort parameter, so you can trade depth of thinking for speed and cost per request.

Where it fits:

• Bulk processing — classifying tickets, tagging content, extracting fields from text into JSON, thousands of similar requests where the unit price matters.
• Routing — a cheap first pass that decides which requests need a bigger model.
• Summaries and compaction — logs, long threads, documents.
• Subagents — the agent that reads files, searches and reports back while Sonnet or Opus makes the decisions.

Where to step up: multi-step refactors, unfamiliar codebases and anything where a wrong turn is expensive belong to Sonnet 5.5 or Opus 5.5.

Haiku 5.5 in the Claude lineup

Against Haiku 4.5 the list price is ten times lower, the context window is five times larger, and effort control is new.

Haiku 5.5 or GPT-6 Luna

The two small models now share a list price: $0.10 per million input and $0.50 per million output tokens. The practical differences:

If your stack is Claude Code, Haiku 5.5 slots in as the background model without changing vendors; if you run Codex, Luna does the same job there. Both work with the same TeamoRouter key — see the GPT-6 Luna article.

Pricing

Per 1 million tokens, in USD:

The live number is always on the pricing page; it moves with the upstream price list, so the figures here are as of the publish date.

What that looks like: classifying 10,000 support tickets of about 1,000 tokens each with a 100-token answer is 10 million input and 1 million output tokens — $1.50 at list price, about $0.44 on TeamoRouter.

One thing that changes the bill compared with Haiku 4.5: the tokenizer is newer, and the same text counts as about 30% more tokens, according to Anthropic.

Claude Code: Haiku 5.5 as the background and subagent model

Claude Code uses the "haiku" slot for background chores such as titles and summaries. Pointing it at Haiku 5.5 makes those cheaper and faster.

First, update Claude Code. Version 2.1.293 (October 7, 2026) is the first that knows Haiku 5.5. We tested 2.1.292 and older: they warn that "claude-haiku-5-5" isn't described by this version's model catalog and treat the context as 200k tokens. Run claude update, then check claude --version.

macOS and Linux — add to ~/.zshrc (or ~/.bashrc):

export ANTHROPIC_BASE_URL=https://api.teamorouter.com
export ANTHROPIC_AUTH_TOKEN="your TeamoRouter key"
export ANTHROPIC_MODEL=claude-sonnet-5-5
export ANTHROPIC_DEFAULT_SONNET_MODEL=claude-sonnet-5-5
export ANTHROPIC_DEFAULT_OPUS_MODEL=claude-opus-5-5
export ANTHROPIC_DEFAULT_HAIKU_MODEL=claude-haiku-5-5

Windows (PowerShell):

setx ANTHROPIC_BASE_URL "https://api.teamorouter.com"
setx ANTHROPIC_AUTH_TOKEN "your TeamoRouter key"
setx ANTHROPIC_MODEL "claude-sonnet-5-5"
setx ANTHROPIC_DEFAULT_SONNET_MODEL "claude-sonnet-5-5"
setx ANTHROPIC_DEFAULT_OPUS_MODEL "claude-opus-5-5"
setx ANTHROPIC_DEFAULT_HAIKU_MODEL "claude-haiku-5-5"

Then reopen the terminal and run claude.

Subagents. To run Claude Code's subagents on Haiku 5.5 as well, add CLAUDE_CODE_SUBAGENT_MODEL=claude-haiku-5-5. This suits subagents that read and search; if yours write code, keep them on Sonnet 5.5.

You can also switch the main session to Haiku with /model → haiku for a batch of simple edits.

Calling the API directly

Anthropic format:

curl https://api.teamorouter.com/v1/messages \
  -H "x-api-key: YOUR_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-haiku-5-5",
    "max_tokens": 512,
    "messages": [{"role": "user", "content": "Classify this ticket as billing, bug or question: ..."}]
  }'

OpenAI-compatible format:

from openai import OpenAI

client = OpenAI(base_url="https://api.teamorouter.com/v1", api_key="YOUR_KEY")
resp = client.chat.completions.create(
    model="claude-haiku-5-5",
    messages=[{"role": "user", "content": "Extract company name, amount and due date as JSON: ..."}],
)
print(resp.choices[0].message.content)

Full parameter reference: API integration.

Moving from Haiku 4.5

• Change the model name to claude-haiku-5-5.
• Remove temperature, top_p and top_k if you set them to non-default values: Haiku 5.5 returns a 400 error for them.
• Thinking is adaptive and on by default; tune it with effort rather than a thinking budget.
• Re-check costs: the newer tokenizer counts more tokens for the same text.

Paying for usage

You top up a balance in the console and are charged only for tokens actually used — no subscription.

• Card — Visa, Mastercard and other major cards via Stripe.
• USDT — "Pay with crypto" in the top-up dialog; credited within one to five minutes.

FAQ

Is claude-haiku-5-5 the same as claude-haiku-4-5? No — different models and IDs. Nothing switches automatically; update your settings.

Why does Claude Code warn about the model catalog? Your Claude Code is older than 2.1.293. Run claude update.

Does Haiku 5.5 have the full 1M context? Yes — 1,000,000 tokens of context and up to 128,000 output tokens, the same as Sonnet 5.5 and Opus 5.5.

Are there request limits? Yes, per model — see the rate limits page.

Next steps

• Claude Sonnet 5.5: pricing and setup — the default model for everyday coding.
• Billing FAQ — how caching is charged.
• Create an API key.