Home · All models · pricing.json

Prices per call

Reference price per call for each model, for one user message of about 2,000 characters and max_tokens 512, compared with the same request on BlockRun. Paid in USDC on Base with x402, no API key.

Prices follow the live supplier cost, so every real request is quoted on its own: the HTTP 402 answer states the exact price before you pay, and a request that fails is refunded. Reference prices updated 2026-10-15.

ModelPriceBlockRunDifferenceQuoted
Claude Opus 5.5
/llm/claude-opus-5.5
$0.0056$0.0059-5.8%2026-10-15
Claude Sonnet 5.5
/llm/claude-sonnet-5.5
$0.0038$0.0035+9.4%2026-10-15
Claude Haiku 5.5
/llm/claude-haiku-5.5
$0.0010-2026-10-15
GPT-6 Astra
/llm/gpt-6-astra
$0.0111$0.0134-16.9%2026-10-15
GPT-6.1 Sol
/llm/gpt-6.1-sol
$0.0038$0.0035+9.4%2026-10-15
GPT-5.5
/llm/gpt-5.5
$0.0070$0.0074-5.9%2026-10-15
GPT-6 Luna
/llm/gpt-6-luna
$0.0014$0.0020-30%2026-10-15
Gemini 4 Argon
/llm/gemini-4-argon
$0.0107-2026-10-15
Gemini 3.1 Pro
/llm/gemini-3.1-pro
$0.0037$0.0036+3.5%2026-10-15
Gemini 3.8 Flash
/llm/gemini-3.8-flash
$0.0023$0.0020+15%2026-10-15
Grok 4.7
/llm/grok-4.7
$0.0036$0.0033+10.3%2026-10-15
Kimi K3
/llm/kimi-k3
$0.0044$0.0047-6.5%2026-10-15
GLM-5.3
/llm/glm-5.3
$0.0029$0.0026+11.7%2026-10-15
DeepSeek V4 Pro
/llm/deepseek-v4-pro
$0.0028$0.0025+12.2%2026-10-15
DeepSeek V4.1 Flash
/llm/deepseek-v4.1-flash
$0.0019-2026-10-15
DeepSeek V4 Flash Chat
/llm/deepseek-chat
$0.0014$0.0020-30%2026-10-15
Qwen3.8 Max
/llm/qwen3.8-max
$0.0036$0.0033+10.3%2026-10-15
MiniMax M3
/llm/minimax-m3
$0.0016$0.0020-20%2026-10-15
Llama 3.3 70B
/llm/llama-3.3-70b
$0.0043-2026-10-15
Mistral Large 4
/llm/mistral-large-4
$0.0023$0.0020+15%2026-10-15
OpenAI GPT OSS 120B
/llm/gpt-oss-120b
$0.0010-2026-10-15
Claude Opus 5
/llm/claude-opus-5
$0.0068$0.0072-5.3%2026-10-15
Claude Sonnet 5
/llm/claude-sonnet-5
$0.0038$0.0035+9.4%2026-10-15
Gemini 3.6 Flash
/llm/gemini-3.6-flash
$0.0023$0.0020+15%2026-10-15

Payment: x402 v2 (scheme exact) in USDC on Base mainnet, one payment per request. Longer prompts and higher max_tokens cost more; check the 402 quote.