Bonus

Chinese models Kimi-k3, Qwen3.8-max, and GLM-5.3 are now available at 50% off their official prices; GLM-5.3-flash and DeepSeek-V4.1-flash-1017 are available at 40% off their official prices.

Login
Models/Text/Claude Haiku 4.5

Anthropic: Claude Haiku 4.5

Reasoning / thinking
Supported input types: Text / Images
Supported output types: Text
Supported by some available channels. Routing filters by the requested capabilities.

Balanced model lane with live routing fallback and production-friendly defaults.

Capability support
TText🖼Vision⚙Tools{}JSON⚡Streaming
Context window128Ktokens

Official pricing

The model's published list prices, used only as the comparison baseline for route pricing and discounts

Input tokens$1.001M
Output tokens$5.001M
Cache read$0.101M
Cache write$1.251M

Available routes

Provider, public group, live pricing, latency, and uptime per active route

Provider / groupInput tokens1MOutput tokens1MCache read1MCache write1MLatencyThroughputUptime
AnthropicGroup: Haiku70% OFF
Input tokens$0.30$1.00
Output tokens$1.50$5.00
Cache read$0.0300$0.10
Cache write$0.38$1.25
Latency1.0sThroughput—
Uptime