// Supported Models

Every major provider.
One platform.

Anthropic, OpenAI, Google, xAI, DeepSeek, Meta, Mistral, Alibaba — plus 300+ more accessible through Nyquest's upstream gateway. Use models in the app, via the public API, or with your own provider keys.

// Two Ways to Use Them

In the app, or programmatically

Same model catalog, two access paths. Pick whichever fits the work.

In the App
Sign in at app.nyquest.ai. Use the model picker to select any of 300+ models. Pin a model to a conversation, or let the dispatcher route automatically based on tier and complexity. All chassis, memory, web grounding, and agents available.
Open App →
Public REST API
Issue an nq-v1-... key from your account settings. Hit https://api.nyquest.ai/v1/chat/completions with any OpenAI-compatible client. Stateless mode (no memory recall, no conversation persistence) — clean developer-API behavior.
API details →
// Provider Catalog

Models available today

A snapshot of headline models. The full catalog is searchable in the app's model picker and at GET /v1/models.

Anthropic

FRONTIER
Claude family — strong at reasoning, instruction-following, and long-context work. Default at every dispatcher tier.
claude-opus-4-7 · T3 / Pro flagship
claude-sonnet-4-6 · T2 / Pro default
claude-haiku-4-5 · T1 / Pro cheap path

OpenAI

GENERAL
GPT family — broad capability, strong tool-use, and competitive pricing across the size range.
gpt-5-5 · flagship
gpt-5-5-pro · pro tier
gpt-5-4-mini · fast + cheap
gpt-5-4-nano · ultra-budget

Google

LONG CONTEXT
Gemini family — million-token context windows, multimodal native, fast inference on Flash variants.
gemini-3-1-pro · 1M context
gemini-3-flash · low-latency
gemini-2-5-flash · budget

xAI

GENERAL
Grok family — 131K context, OpenAI-compatible format, strong on real-time reasoning tasks.
grok-3 · flagship
grok-3-mini · cost-efficient

DeepSeek

REASONING
DeepSeek family — strong at math and reasoning at low cost. Used internally by the engine for semantic compression.
deepseek-r1 · reasoning-tuned
deepseek-v3-2 · general-purpose

Open Weights

300+ MORE
Meta Llama, Mistral, Qwen, Gemma — accessible through the same picker. Search by capability, size, or context length in the app.
llama-4-maverick
mistral-large
qwen3-235b · and many more

Model identifiers above are illustrative. The live catalog syncs daily — query GET /v1/models for canonical IDs and current availability.

// How It Works

Two paths to the model

Hosted requests go through Nyquest's upstream gateway. BYOK requests skip it and hit the provider directly using your key.

Hosted (wallet-billed)
You │ ▼ Nyquest │ tier select · compress · dispatch ▼ Upstream Gateway │ ▼ Provider API

Wallet draws at standard provider rates — no markup. Compression and routing happen before dispatch.

BYOK (your key)
You │ ▼ Nyquest │ decrypt your key · compress ▼ Provider API (direct) │ billed to YOUR account ▼ You

Direct call to the provider using your key. No upstream gateway. Your rates, your billing. Keys AES-256-GCM encrypted at rest.

// Model Selection

Auto-routed or pinned — your choice

When you don't specify, the dispatcher picks based on your tier and the request's complexity. When you do specify, that model handles every message in the conversation.

Tier-based defaults
Free accounts route to whichever models are currently available at no cost. Once you fund your wallet, Pro unlocks every tier — T1 (Haiku 4.5), T2 (Sonnet 4.6), T3 (Opus 4.7), and the full 369+ model catalog. Circuit breakers fail over between tiers if a provider has an outage.
Pinning
Pick any model from the picker and pin it. That model handles the whole thread — no routing decisions, no surprises. Great for benchmarking or when you have a strong preference.
Visibility
Every request shows which model served it, how many tokens were used, and what it cost. Available in usage analytics and on the request inspect panel.

Try a model

Sign up free with BYOK or any model currently available at no cost. Switch models any time from the picker, or pin one for the whole thread.

Open Nyquest →