by ~anthropicReleased 27 April 2026
Haiku 5.5 is the quick, small Claude for high-volume work: summarising, sorting, pulling out details, routing, driving a web browser and acting as a helper agent. It reads text, images and files, keeps about 1,000 pages in mind, replies up to about 128 pages, and has an effort setting while deciding for itself how long to think. It succeeds Haiku 4.5, with better coding and computer use.
You pay for what each request uses. No subscription, no minimum.
One message, the ordinary use
Every rate
| Meter | ↑ you send | ↓ comes back |
|---|---|---|
| Text tokens | 33,28 DA | 166,42 DA |
| Cached tokens read | 3,33 DA | |
| Cache-write tokens | 41,60 DA |
per 1M tokens
Web search · 3,33 DA per call
One company runs this model. OpenDunes sends every request to it. DA per 1M tokens: ↑ what you send · ↓ what comes back.
| Provider | ↑ Sent | ↓ Back | Latency | Throughput | Uptime |
|---|---|---|---|---|---|
| Anthropic: Claude Haiku Latest | 33,28 DA | 166,42 DA | … | … | … |
Also charged: cache write ↑ 41,60 DA per 1M tokens · web search 3,33 DA per search.
You pay the price of the provider that served your request. Speed over the last 30 minutes · uptime over 3 days.
Real requests, real replies. Try any of them yourself — one click, pre-filled.
What's the total on this invoice, and when is it due?
invoice-scan.jpg · photoThe total is 42 300,00 DA, due 5 September 2026. Worth a double-check: the handwritten day could read as 8 instead of 3.
Try in Chat LabTurn this customer email into JSON with intent, urgency and order_id.
customer-email.txtTry your own request — same model, pre-selected.
Open Chat LabSwap the base URL and the key. Your OpenAI code runs unchanged.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://opendunes.com/v1",
api_key=os.environ["OPENDUNES_API_KEY"],
)
resp = client.chat.completions.create(
model="~anthropic/claude-haiku-latest",
messages=[{"role": "user", "content": "Two lines to welcome a shop in Oran."}],
)
print(resp.choices[0].message.content)Runs this request once on your balance.
Standard OpenAI fields, grouped by what they control. Anything not listed is accepted and ignored.
Shaping the reply
How long, how varied, where to stop.
max_tokensstopReasoning
Let it think before answering — and see the thinking.
reasoningreasoning_effortinclude_reasoningStructured output
Get JSON you can parse instead of prose.
response_formatstructured_outputsTools
Let the model call your functions.
toolstool_choiceOther
Also accepted by this model.
max_completion_tokensverbosityAbout 0,12 DA per message. ↑ 33,28 DA · ↓ 166,42 DA per 1M tokens.
No. You pay per use from your balance, with no subscription. Try it in Chat at the same price.
Haiku 5.5 is the quick, small Claude for high-volume work: summarising, sorting, pulling out details, routing, driving a web browser and acting as a helper agent. It reads text, images and files, keeps about 1,000 pages in mind, replies up to about 128 pages, and has an effort setting while deciding for itself how long to think. It succeeds Haiku 4.5, with better coding and computer use.
Deposit dinars with your Edahabia or CIB card. Every model, Claude Haiku Latest included, is paid from that balance. No foreign card, no conversion.
Anthropic reports Claude's results in 14 languages besides English, Arabic and French among them, and says Claude works in many more. Darja is not named.
It reads text, images and PDFs. Up to 1M tokens in one request, about 1 000 pages. It writes up to 96 000 words at once.
No. ~anthropic has not published its weights.
~anthropic. Released 27 April 2026.
#1 for coding · Claude Fable 5.1 reads text, images and files, keeps about 1,000 pages in mind at once, and replies in text up to about 128 pages. Give it demanding coding work, big changes across a whole codebase, interface and visual code, or finance and analysis work that runs many steps. Its plans and summaries come out shorter than Fable 5's. Anthropic advises comparing it with Opus 5.5 on hard reasoning; Opus 5.5 is the default start, and Sonnet 5.5 suits everyday jobs.
#1 for reasoning · Claude Opus 5.5 reads text, images and files, keeps about 1,000 pages in mind at once, and replies up to about 128 pages, with an effort setting as the main dial between depth and speed. Give it edits that span a large codebase, bug hunts, financial or scientific analysis, and charts, diagrams or screenshots full of detail. It needs fewer steps and less output than Opus 5 for the same job and reports back in plainer language. For harder reasoning, Anthropic points to Fable 5.1; for lighter work, Sonnet 5.5.
#2 for interface design · Claude Sonnet 5.5 reads text, images and files, keeps about 1,000 pages in mind at once, and replies up to about 128 pages, with thinking always on and an effort setting as the main dial. It is Anthropic's pick for well-defined daily jobs: shipping a feature, fixing a bug, or turning notes into finished reports, slide decks and spreadsheets. Lower effort keeps it responsive in agents that work step by step. For high-volume sorting and extraction, Haiku 5.5 is faster; for long, hard projects, Opus 5.5.
From what its maker and two benchmark sites publish, and where it stands on OpenDunes. Read 7 Nov 2026.
Thinks first
Works through the problem before it answers.
Uses your tools
Can call functions your code provides.
Fills your JSON
Answers in the JSON shape you give it.
Searches the web
Can look things up online while it answers.
Cheaper on repeats
A prompt it saw a moment ago costs less the second time.
Open weights No.
Read on platform.claude.com.
Reads a million tokens of documents, images or files and returns a tight summary
Calls your tools and fills your JSON for subagent loops running at high volume
Uses your tools · Fills your JSON
Thinks through a coding fix before writing it, keeping the reply short and quick
Thinks first