Corti Models

Frontier AI models on sovereign infrastructure, through an OpenAI-compatible API. Use them in your coding agent or call them directly.

Corti Models gives you frontier AI performance without giving up sovereignty, control, or speed. Corti takes the strongest available open-weight models and turns them into production-grade AI through tuning, inference optimization, and continuous improvement.

Why Corti

• Faster: optimized for low latency (180 ms TTFT, 118.5 tokens/s on corti-s1)
• Cheaper: up to 50% lower cost than equivalent frontier models
• Sovereign: inference runs on Corti-managed infrastructure, never touching third-party cloud providers in the request path
• Zero retention: prompts and completions are never stored, never logged, and never used for training
• Always frontier: Corti continuously upgrades the underlying model behind each stable ID, so you always get the best available model without changing your code

For EU projects, inference runs entirely on Corti-managed European infrastructure with zero data retention. US projects are served by Azure-hosted OpenAI models under the same model IDs. See Regions.

Models

Corti S1 and S1 Mini are heavily optimized foundation models. Corti continuously selects the most capable model for each price tier and tunes it for quality and latency, so you always get the best available performance behind a stable model ID.

S1 is a product, not just a model. See How Corti models work for what’s behind the model IDs.

Corti S1

• Based on GLM 5.2
• Frontier quality (53 on AA Intelligence Index)
• Optimized for latency (180 ms TTFT and 118.5 tokens/s)

Corti S1 Mini

• Based on Qwen 3.6
• Efficient quality (31 on AA Intelligence Index)
• Supports image input

Choose your path

Use Corti Models in your coding agent

Point OpenCode, Pi, Crush, or ForgeCode at Corti Models with the Corti CLI. No code changes.

Replace your LLM provider

Swap your base URL and API key. Your existing OpenAI SDK code works unchanged.

Sovereign deployment

Corti operates zero data retention: prompts and completions are never stored, never logged, and never used for training. For EU projects, inference runs on Corti-managed European infrastructure and data never leaves Europe. For US projects, the same model IDs are served by Azure-hosted OpenAI models. The platform can also be deployed in a dedicated environment or on customer-controlled infrastructure including on-premises. Contact us to discuss deployment options.

Next steps

Models

Compare all four model variants by capability, reasoning, speed, and price.

How Corti models work

Learn how Corti turns open-weight models into production-grade AI.

Regions

See what’s served where, and how to reach each region.

API Reference

Browse the full Corti Models API with interactive examples.