SoxAI

About SoxAI

SoxAI is an AI API gateway unifying 200+ models from 6 providers behind a single OpenAI-compatible endpoint — built for teams that need per-developer API keys, org-level quota policies, and automatic failover.

What SoxAI is

A single HTTP endpoint (api.soxai.io) that speaks the OpenAI, Anthropic Messages, and Responses APIs, routes each request to the most appropriate upstream provider, handles retries and failover automatically, and bills you with itemized per-developer usage.

  • API compatibility: OpenAI Chat Completions, Anthropic Messages, Responses (for Codex CLI), image-generation endpoints.
  • Providers covered: Anthropic (Claude 4.x), OpenAI (GPT-5.x + image), Google (Gemini 2.5 / 3.x + multimodal), GLM (Zhipu), MiniMax, Gemma open weights.
  • Catalog size: 61 models in the public catalog as of 2026-04-17; 200+ including BYOK channels. See the supported models list.
  • Deployment: Hosted at api.soxai.io / console.soxai.io, with a source-available self-host option for teams that need data residency, air-gapped operation, or white-label.

At a glance

API surfaceOpenAI-compatible HTTP
Models supported200+ (61 in public catalog)
Providers6 — Anthropic, OpenAI, Google, GLM, MiniMax, Gemma
Free tier$0.50 trial credit, no card required
First top-up bonus$5 bonus on first top-up ≥ $5
BillingPer-token, USD, itemized invoices
Gateway latencyGateway P50 target under 5 ms
Throughput3,000 RPS + 2,000 concurrent streams (4-core baseline)
SecurityAES-256-GCM for keys, Argon2id passwords, WebAuthn MFA, Row-Level Security
Support languagesEnglish, Chinese
Contact[email protected]

Who SoxAI is for

Why we built SoxAI

Every engineering team we talked to was running into the same three walls at once: provider rate limits breaking production during peak hours, shared API keys making spend impossible to attribute per developer, and no way to set a hard cap before someone's runaway loop ate the monthly budget. SoxAI exists to fix all three — without requiring an enterprise contract or a DevOps rebuild.

Get in touch

Have questions, want a demo, or need to discuss self-host / white-label deployment? Email us directly.