Claude Haiku API in India

Claude Haiku is Anthropic's fastest, smallest Claude model, and on unoblox it's available from India through one ₹-billed, OpenAI-compatible endpoint — the same key and base URL you'd use to call Claude Sonnet, Claude Opus, or any other model in the catalogue, settled on a single monthly GST invoice.

What Claude Haiku is built for

Haiku trades some of the depth of Claude Opus and Claude Sonnet for speed and lower per-token cost, which makes it a fit for latency-sensitive chat, high-volume classification, content moderation passes, and any step in a pipeline that runs on every request. Because it shares the same Claude API shape as Sonnet and Opus, a prompt written against Haiku generally ports to a larger Claude model with little more than a model-id change if quality needs to go up for a subset of traffic.

₹ pricing: Haiku vs Sonnet vs Opus

Claude Haiku's live ₹ rate is on its model page rather than fixed in this note. The rest of the Claude line is priced as follows:

Model	Input (₹ / 1M tokens)	Output (₹ / 1M tokens)
Claude Sonnet	201.6	1008
Claude Opus	504	2520

As the smallest tier, Haiku is priced below Sonnet — worth confirming the exact current figure on the model page before sizing a high-volume workload.

Calling Claude Haiku from the unoblox endpoint

curl https://api.unoblox.ai/v1/chat/completions \
  -H "Authorization: Bearer ub-gw-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-haiku-4-5",
    "messages": [{"role": "user", "content": "Tag this support message with a one-word category."}]
  }'

The same request format, streaming and tool-calling support carry over unchanged if you swap model to anthropic/claude-sonnet-5 or anthropic/claude-opus-5 for a harder subset of requests.

Choosing Haiku over Sonnet or Opus

• Claude Haiku — highest volume, latency-sensitive, cost-per-call matters most.
• Claude Sonnet — the usual default for production features needing stronger reasoning than Haiku offers.
• Claude Opus — reserved for the hardest requests, where getting the answer right outweighs cost and latency.

A common setup routes most traffic to Haiku and escalates only ambiguous or high-value requests to Sonnet or Opus.

Frequently asked questions

Is Claude Haiku available in India through unoblox? Yes, anthropic/claude-haiku-4-5 is callable on https://api.unoblox.ai/v1 immediately after signup, billed in rupees.

What does Claude Haiku cost on unoblox? It isn't in our fixed indicative price list; the live ₹ rate is on its model page and sits below Claude Sonnet's ₹201.6 / ₹1008 per 1M tokens.

Can I use Claude Haiku and Claude Opus with the same API key? Yes, one ub-gw- key works across every Claude model and every other model in the catalogue.

Does using Claude Haiku mean data is processed in India? No, it runs on Anthropic's infrastructure. The India-specific benefit from unoblox is ₹ billing, a GST invoice and one endpoint, not data residency.

Is Claude Haiku good for high-volume chat features? Yes, its lower latency and cost per call make it a common choice for high-frequency, lower-complexity conversational steps.

Is a GST invoice issued for Claude Haiku usage? Yes, combined with all other model usage into one monthly GST invoice from an Indian entity, input tax credit claimable.

Get started in rupees → https://unoblox.ai/sign-in