APIHub

Model pricing

All prices in USD per 1M tokens, billed separately for input and output. Struck-through figures are official list prices.

11 models

ModelCapabilitiesContextInput / 1MOutput / 1M
Claude Opus 56%
claude-opus-5
VisionTool useReasoningLong context
200K
$0.840
$15.00
$4.20
$75.00
Claude Sonnet 58%
claude-sonnet-5
VisionTool useReasoningLong context
200K
$0.240
$3.00
$1.20
$15.00
Claude Opus 4.86%
claude-opus-4-8
VisionTool useReasoningLong context
200K
$0.840
$15.00
$4.20
$75.00
Claude Haiku 4.512%
claude-haiku-4-5
VisionTool use
200K
$0.120
$1.00
$0.600
$5.00
GPT-5.210%
gpt-5.2
VisionTool useReasoningLong context
400K
$0.120
$1.25
$0.960
$10.00
GPT-5 mini16%
gpt-5-mini
VisionTool use
400K
$0.040
$0.250
$0.320
$2.00
Gemini 3 Pro10%
gemini-3-pro
VisionTool useReasoningLong context
1M
$0.120
$1.25
$0.960
$10.00
Gemini 3 Flash20%
gemini-3-flash
VisionTool useLong context
1M
$0.015
$0.075
$0.060
$0.300
Grok 410%
grok-4
VisionTool useReasoning
256K
$0.288
$3.00
$1.44
$15.00
DeepSeek V324%
deepseek-v3
Tool use
128K
$0.065
$0.270
$0.264
$1.10
DeepSeek R124%
deepseek-r1
ReasoningTool use
128K
$0.132
$0.550
$0.526
$2.19

How billing works

Billed per token

After each request we charge the exact input and output token counts reported upstream. No rounding up, no packaged blocks.

Failed requests are free

Anything returning 4xx or 5xx costs nothing. If a stream drops midway you are only charged for tokens actually generated.

Credit never expires

Top-ups have no expiry date, no monthly fee and no minimum spend. Subscription allowances fall back to pay-as-you-go once exhausted.

Price changes announced ahead

If an upstream provider changes pricing we announce it on-site and update this page first. Credit you already hold is unaffected.