Model library

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

Fast, low-cost Claude for high-volume tasks.

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

Anthropic's strongest model for long-horizon agentic and coding work.

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

Balanced Claude for everyday coding, analysis and tool use.

Codestral by Mistral, served through the NovaServe gateway.

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

DeepSeek's chain-of-thought reasoning tier.

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

Cost-efficient open-weight general model.

Devstral Medium by Mistral, served through the NovaServe gateway.

Balanced Gemini with lower cost and latency than Pro.

Cheapest Gemini tier for simple, high-volume work.

Large-context multimodal reasoning across text, image, audio and video.

Preview Flash model balancing speed and capability.

Cost-efficient Gemini for classification and summarization.

Google's strongest reasoning model, with very large context.

High-efficiency Gemini for fast agentic workflows.

Latest Gemini Flash — fast coding, reasoning and agent loops.

Newest Gemini speech preview.

Gemini speech synthesis with prebuilt voices.

Lightweight Gemini speech for short clips.

Highest-quality Gemini speech synthesis.

Smaller, cheaper OpenAI image model for drafts.

OpenAI image model — strong at legible text and typography.

General-purpose all-rounder with text and image input.

Lower-cost GPT-5 tier for general workloads.

Fastest GPT-5 tier for simple, high-volume calls.

Prior-generation reasoning model, still strong on multi-step problems.

Frontier-class coding and analysis at a lower price point.

Strong mini model for sub-agents and high-volume coding work.

Cheapest GPT-5.4 tier — classification, extraction and ranking.

Premium reasoning variant for the most complex tasks.

Frontier model for complex coding, analysis and professional work.

Extended-reasoning variant for the hardest problems.

Fast, low-cost model for short replies and high-volume traffic.

Flagship reasoning model — the default behind RouteNova for hard tasks.

Balanced everyday model: near-flagship quality at lower cost.

GPT-OSS 120B by Groq, served through the NovaServe gateway.

GPT-OSS 20B by Groq, served through the NovaServe gateway.

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

xAI's frontier reasoning model with real-time knowledge.

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

Low-latency Grok tier for agent loops and tool calling.

Awaiting provider funding. This model is wired up but its provider account needs funding before it can be called. Requests are refused until our team tops it up.

Grok tuned for fast code generation and edits.

Groq Compound by Groq, served through the NovaServe gateway.

Groq Compound Mini by Groq, served through the NovaServe gateway.

Magistral Small by Mistral, served through the NovaServe gateway.

Ministral 3B by Mistral, served through the NovaServe gateway.

Ministral 8B by Mistral, served through the NovaServe gateway.

Mistral Large by Mistral, served through the NovaServe gateway.

Mistral Medium by Mistral, served through the NovaServe gateway.

Mistral Small by Mistral, served through the NovaServe gateway.

Quick drafts and iterations at the lowest image price.

Fast image generation and editing at pro-level quality.

Highest-fidelity image generation and editing.

Accurate speech-to-text with punctuation and formatting.

Faster, cheaper transcription for bulk audio.

Text-to-speech with steerable delivery and adjustable speed.

Qwen 3.6 27B by Groq, served through the NovaServe gateway.

Highest-quality video generation with audio, up to 8s.

Faster Veo tier for iteration at moderate cost.

Cheapest Veo tier — default for drafts and previews.