All models
Anthropic · text
Claude Haiku 4.5
claude-haiku-4-5
Low-latency Claude route for high-volume workloads.
Pricing
Input$0.35/M
Output$1.76/M
−65%vs. official price
Billed 1:1 against prepaid credits — no markup, no subscription.
Tokens (30d)
10.19M
processed via gateway
Throughput
135 tok/s
typical output speed
Latency
644ms
p95 1.03s
Success
99.9%
2xx responses
Usage & performance
Representative figures based on recent gateway activity.
Token throughput
Last 30 daysLatency
Last 30 daysSpecifications
Model IDclaude-haiku-4-5
ProviderAnthropic
ModalityText & code
Endpoint/api/v1/messages
Context window200,000 tokens
Price$0.35/$1.76 per 1M tokens
vs. official~65% cheaper
Supported parameters
messagestemperaturemax_tokensresponse_formattoolstool_choicetop_p
How to use
Point any OpenAI-compatible client at the gateway and call claude-haiku-4-5.
curl https://api.blackmagic.engineering/v1/chat/completions \
-H "Authorization: Bearer $BLACKMAGIC_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-4-5",
"messages": [
{ "role": "user", "content": "Draft a launch note." }
]
}'Same OpenAI SDK — only the base URL changes.