Home

Hydite Vtslx AO

The unified LLM API aggregation and orchestration platform

Hydite Vtslx AO is a unified API aggregation platform that brings together hundreds of large language models (LLMs) from dozens of providers into a single, consistent API interface. Whether you're building AI-powered applications, running model evaluations, or operating an enterprise-scale AI infrastructure, Vtslx AO eliminates the complexity of managing multiple provider SDKs, API keys, and billing relationships.

Supported models and providers#

OpenAIAnthropicGoogleMetaDeepSeekQwenZhipuMoonshotMiniMaxByteDanceStepFunxAIMistralCohere

Vtslx AO aggregates models across the entire LLM ecosystem:

  • Global frontier models — OpenAI (GPT-5.5, GPT-5.5 Pro, GPT-5.2, GPT-5.5 Instant), Anthropic (Claude Opus 4.7, Claude Sonnet 4.6, Claude Haiku 4.5, Claude Mythos), Google (Gemini 3.1, Gemini 3.1 Flash), Meta (Llama 4 Behemoth, Llama 4 Maverick, Muse Spark)
  • Chinese domestic models — DeepSeek (V4 Pro, V4 Flash, R1-0528), Qwen (Qwen3.6, Qwen3.6-Max, Qwen3-Coder-Next, QwQ-32B), Zhipu (GLM-5), Moonshot (Kimi K2.6), MiniMax (M2.7), ByteDance (Doubao Seed 2.0), StepFun (Step 3.5 Flash)
  • Open-source and specialized models — Mistral (Large 3, Small 4), xAI (Grok Heavy, Grok 4), Cohere (Command R+), and more

All models are accessible through a single API endpoint with OpenAI-compatible request/response format, meaning you can switch between models by changing one parameter — no SDK changes required.

Deployment options#

Vtslx AO offers two deployment tiers to match your scale and security requirements:

Shared instance (Edge API)#

Shared instances are deployed across globally distributed edge nodes, providing low-latency access from any region. This tier is ideal for:

  • Development and testing
  • Small to medium-scale production applications
  • Projects that don't require data residency guarantees

Shared instances share compute resources with other tenants, with fair-use rate limiting to ensure consistent performance.

Dedicated instance (Enterprise)#

Dedicated instances provision isolated compute nodes exclusively for your organization. This tier delivers:

  • Higher concurrency — No shared throttling; burst to your licensed capacity
  • Faster response times — Dedicated GPU/CPU resources eliminate noisy-neighbor latency
  • Private model deployment — Deploy fine-tuned or proprietary models directly to your dedicated nodes
  • Data residency — Choose deployment regions that meet your compliance requirements
  • Custom domain — Serve API traffic through your own domain with full TLS control
  • SLA guarantees — Enterprise-grade uptime commitments and priority support

Key features#

Unified API format#

All models speak the same language. Vtslx AO exposes an OpenAI-compatible /v1/chat/completions endpoint, so any existing OpenAI SDK or tooling works out of the box with every supported model.

API key groups#

Organize API keys into logical groups with shared configuration — rate limits, budget caps, allowed model lists, and system prompts. Groups make it easy to manage access for different teams, environments, or customer tiers.

Fine-grained rate limiting#

Control usage at both the group and individual key level:

  • RPM (requests per minute)
  • TPM (tokens per minute)
  • Budget caps (daily, weekly, monthly, or yearly spend limits)

System prompts at the gateway#

Inject system prompts at the API gateway level, ensuring consistent behavior across all consumers of a key group without requiring client-side changes.

Usage analytics#

Track token consumption, request volumes, latency distributions, and cost breakdowns by model, key, and group — all from the Vtslx AO dashboard.

Getting started#

  1. Create an organization in the Vtslx AO Dashboard
  2. Create an API key group with your desired rate limits and model access
  3. Generate an API key and start making requests:
1
curl https://api.hydite.com/v1/chat/completions \
2
  -H "Content-Type: application/json" \
3
  -H "Authorization: Bearer YOUR_API_KEY" \
4
  -d '{
5
    "model": "gpt-5.5",
6
    "messages": [{"role": "user", "content": "Hello!"}]
7
  }'

To switch models, just change the model field — no other code changes needed.