At a glance
- Routed as
- claude/claude-haiku-4-5-20251001
- Context window
- 200,000 tokens
- Max output
- 64,000 tokens
- Reasoning format
- claude-budget · can be disabled
Price per million tokens
Input
$1
Output
$5
Cached input
$0.1
Published by the provider. ShareQuota adds nothing to it — there is no markup and no fee, because there is nothing in between to charge for.
Capabilities in full
Both columns are shown. What a model cannot do is a fact about it, and hiding it just moves the surprise to runtime.
- Reasoning
- Vision
- Tools
- Web search
- PDF input
- Audio in
- Video in
- Image out
- Audio out
These models are served through a claude.ai subscription login. That credential is only accepted by Claude Code and Anthropic's own clients — for third-party SDKs, connect an Anthropic API key instead.
Call it
The model id below is what goes in the request. Any client that speaks one of the four dialects can send it — the proxy resolves the provider and translates the shape.
curl http://localhost:20130/v1/messages \ -H "x-api-key: sq-your-internal-key" \ -H "content-type: application/json" \ -d '{ "model": "claude/claude-haiku-4-5-20251001", "max_tokens": 256, "messages": [{"role": "user", "content": "Hello"}] }'