Inference Marketplace

Production LLM inference — frontier and open models, no subscription. Pay per request in USDC, or run on a prepaid token balance through an OpenAI-compatible API.

6
Models live
5
Providers
$0.0014
Starting from
Solana · x402
Settlement
Claude Fable 5
Anthropic
$0.05–$0.41
VisionSystem prompt
Claude Fable 5 text generation (Anthropic). Dynamic pricing: estimated input tokens at $0.01/1K + output tier ceiling at $0.05/1K. Tiers: small (1K output tokens), medium (4K), large (8K). Supports system prompt and image input.
POST/llm/claude-fable-5
Claude Opus 4.6
Anthropic
$0.03–$0.20
VisionSystem prompt
Claude Opus 4.6 text generation (Anthropic). Dynamic pricing: estimated input tokens at $0.005/1K + output tier ceiling at $0.025/1K. Tiers: small (1K output tokens), medium (4K), large (8K). Supports system prompt and image input.
POST/llm/claude-opus-4-6
Gemini 3.5 Flash
Google
$0.01–$0.07
System prompt
Gemini 3.5 Flash text generation (Google). Dynamic pricing: estimated input tokens at $0.0015/1K + output tier ceiling at $0.009/1K. Tiers: small (1K output tokens), medium (4K), large (8K). Supports system prompt.
POST/llm/gemini-3.5-flash
GPT-5.2
OpenAI
$0.01–$0.11
VisionReasoning
GPT-5.2 text generation (OpenAI). Dynamic pricing: estimated input tokens at $0.00175/1K + output tier ceiling at $0.014/1K. Tiers: small (1K output tokens), medium (4K), large (8K). Supports image input and reasoning effort control.
POST/llm/gpt-5.2
NVIDIA NIM
NVIDIA
usage-based
System prompt7 modelsOpenAI-compatible
OpenAI-compatible chat completions on frontier models (Claude Fable 5, Gemini 3.5 Flash, GPT-5.2 via Replicate; Nemotron 3 Ultra, Kimi K2.6, DeepSeek V4 Pro, GLM-5.1 via NVIDIA NIM with OpenRouter fallback). Pricing is usage-based: input + output tokens at the provider rate + 10%. On the x402 door the per-call charge is quoted up front as an upper bound (set max_tokens to lower it). Also available drop-in at POST /v1/chat/completions with an API key + prepaid credits (billed exactly per token).
POST/llm/nim
Verified Inference (Prova)
Verified · TEE
$0.0014–$0.01
System prompt12 modelsOn-chain verified
OpenAI-compatible chat completions that run inside a TEE (NVIDIA + Intel), are hardware-attested, and anchored on Solana. Every response returns an on-chain proof + Solscan link. Pick a model via the Model field; price varies by tier (standard / premium / flagship).
POST/llm/prova
💳

How payment works

Every model is gated by the x402 HTTP payment protocol. Call an endpoint and you receive a 402 Payment Required response with payment instructions. Complete the Solana USDC payment, then retry with the payment proof header — your inference runs automatically. No accounts, no API keys, metered per request.