Claude Haiku 4.5

Haiku 4.5 is the quick one. When it came out in October 2025 it matched the coding of the then-current Sonnet 4 at a third of the cost and more than twice the speed, scoring 73.3% on SWE-bench Verified. It was also the first Haiku that could think before answering and operate a computer.

Anthropic pitches it as a worker a bigger model directs — many Haikus running in parallel under one Opus — and for anything a customer waits on, like live chat. It is the oldest model in Anthropic’s current line, and that shows in what it knows.

For scale: 1,000 requests of about 3,000 tokens in and 1,000 out cost $9.20 — a long prompt and a page of answer each. Every rate is published and stays the same from one request to the next. See every model.

Claude Haiku 4.5 is billed in rupees, on the same monthly invoice as every other model your team uses — no foreign card, no separate Anthropic account, and one key for all of it.

Prompts sent to Claude Haiku 4.5 are processed outside Sri Lanka. If a workload has to stay in the country, Qwen3.8-27B (Colombo) runs on our own servers in Colombo.

Anthropic’s published language tests for Haiku cover Hindi (92% of its English score) and Bengali (90%), but not Sinhala or Tamil. Test it on your own messages first.

Use the OpenAI SDK you already have — point it at Roar AI and name the model. Read the docs.

from openai import OpenAI

client = OpenAI(base_url="https://api.roar-ai.com/v1", api_key="roar_live_…")

reply = client.chat.completions.create(
    model="claude-haiku-4-5",
    messages=[{"role": "user", "content": "Summarise this contract in plain English."}],
)