Venice API Overview

Venice.ai is an OpenAI-compatible inference platform for text, image, audio, video, and embeddings. One API — two ways to pay: a traditional API key (Pro account), or a wallet (x402, USDC on Base or Solana, no account required).

Use when

• You're writing code against api.venice.ai for the first time.
• You need to decide between API-key and x402/wallet authentication.
• You want a quick map of which endpoint to call for which task.
• You need to understand the common response headers (X-Balance-Remaining, PAYMENT-REQUIRED, etc.).

Base URL

All endpoints live under:

https://api.venice.ai/api/v1

The OpenAPI spec is distributed at outerface/swagger.yaml (current version 20260728.190807).

Authentication (pick one per request)

Every inference endpoint accepts either — see [venice-auth](../venice-auth/SKILL.md).

# Bearer
curl https://api.venice.ai/api/v1/models \
  -H "Authorization: Bearer $VENICE_API_KEY"

# x402 / SIWE (one-liner via the SDK)
import { VeniceClient } from 'venice-x402-client'
const v = new VeniceClient(process.env.WALLET_KEY)
await v.models.list()

Endpoint map

Inference

Catalog

Account, billing, wallet

Utility

Response headers to watch

Pricing model at a glance

• Pricing is dynamic per request, metered in USD.
• Paid inference endpoints in the spec carry an x-payment-info block with min and max bounds in USD (typically min: 0.001, max: 10.00; higher for bulk video/audio). Read-only discovery routes like GET /models, /models/traits, and /models/compatibility_mapping do not.
• Pro (Bearer) accounts draw from DIEM (staked credits), USD balance, and bundled credits in priority order.
• x402 (wallet) users draw from a prepaid USDC credit balance on Base.
• The authoritative per-model price is on GET /models → model_spec.pricing (when present — video models omit it; use /video/quote for video pricing) (see [venice-models](../venice-models/SKILL.md)).

Standard error shape

Every error body follows one of:

{ "error": "Human-readable message" }

or, for 400 validation errors:

{ "error": "...", "details": { "fieldName": { "_errors": ["Field is required"] } } }

402 on x402 adds structured topUpInstructions and siwxChallenge. See [venice-errors](../venice-errors/SKILL.md) for the full table and retry strategy.

OpenAI compatibility — what works and what doesn't

• Drop-in: /chat/completions, /embeddings, /images/generations, /audio/speech, /audio/transcriptions, /models.
• Ignored but accepted for compat: user, store.
• Venice-only extensions live under:

- veniceparameters (chat completions) - veniceparameters is rejected on /responses — use headers / native fields instead

• Model feature suffixes (e.g. zai-org-glm-5-1:enablewebsearch=on, kimi-k2-6:stripthinkingresponse=true&disablethinking=true) flip veniceparameters via the model ID — see [venice-chat](../venice-chat/SKILL.md#model-feature-suffixes).

Versioning

• info.version in swagger.yaml is a timestamp (YYYYMMDD.HHMMSS). There is no /v2; features roll forward on the single /api/v1 surface and are guarded by:

- Alpha/Beta tags in endpoint descriptions (e.g. /responses, Billing). - x-guidance / model capability flags on /models.

• Always check the model's model_spec.capabilities from GET /models for feature flags (supportsWebSearch, supportsReasoning, supportsE2EE, supportsXSearch, supportsMultipleImages, supportsFunctionCalling, supportsAudioInput, supportsVideoInput, …) before relying on a feature.

Fast start checklist

1. Read [venice-auth](../venice-auth/SKILL.md) and choose Bearer vs x402.
2. GET /models — pick a model and note its modelspec.constraints and modelspec.pricing.
3. Wire up one happy-path call from the matching skill.
4. Add error handling using [venice-errors](../venice-errors/SKILL.md) (402, 422, 429).
5. Hook up observability via X-Balance-Remaining / /billing/usage-history / /x402/transactions.