Skip to main content
Detailed documentation about Corti Models is available here.
ModelsRelease

corti-s1-mini upgraded to Qwen 3.8 (EU)

The efficient tier behind the stable corti-s1-mini and corti-s1-mini-instant model IDs now runs on Qwen 3.8 (previously Qwen 3.6) on Corti’s EU infrastructure. US projects continue to be served by Azure-hosted OpenAI models behind the same IDs (see Regions).
  • Same stable IDs and pricing: keep calling corti-s1-mini and corti-s1-mini-instant.
  • Image input: still supported on both IDs.
  • Reasoning effort: corti-s1-mini now supports reasoning effort (reasoning_effort in Chat Completions, reasoning.effort in the Responses API) with low, medium, or xhigh (default). Other values return a 400 error. To turn reasoning off, use corti-s1-mini-instant. See Models.
  • Benchmarks: AA Intelligence Index 20 (v5.0.1); throughput and time-to-first-token figures will follow.
ModelsRelease

corti-s1 upgraded to GLM 5.3 (EU)

The frontier tier behind the stable corti-s1 and corti-s1-instant model IDs now runs on GLM 5.3 (previously GLM 5.2) on Corti’s EU infrastructure; US projects continue to be served by Azure-hosted OpenAI models behind the same IDs (see Regions). No code changes are needed — model IDs, API surface, and pricing are unchanged.
  • Same stable IDs: keep calling corti-s1 and corti-s1-instant; the foundation model upgrade happens behind the IDs. See How Corti models work.
  • Same API: OpenAI-compatible request and response formats are unchanged.
  • Benchmarks: AA Intelligence Index 45; throughput and time-to-first-token figures will follow on the Models page.
As always, Corti continuously evaluates and may replace the underlying foundation model behind each stable ID as the frontier moves — you always get the best available model without changing your code.
ModelsRelease

Corti Models General Availability General Availability

Corti Models is now generally available. The corti-s1 family — corti-s1, corti-s1-instant, corti-s1-mini, and corti-s1-mini-instant — is served through an OpenAI-compatible API on Corti-managed sovereign infrastructure with zero data retention.
  • Four text-generation variants combining model tier (frontier, efficient) and reasoning mode (with reasoning, instant) — see Models.
  • corti-s1 (frontier tier) is powered by GLM 5.2; corti-s1-mini (efficient tier) is powered by Qwen 3.6. Corti continuously evaluates and may replace the underlying foundation model behind each stable ID as the frontier moves.
  • Embeddings via corti-s1-embedding for semantic search, clustering, and RAG pipelines.
  • Sovereign deployment: EU inference runs entirely on Corti-managed European infrastructure; US projects are served by Azure-hosted OpenAI models under the same model IDs. See Regions.
  • OpenAI-compatible API: swap your base URL and API key — existing OpenAI SDK code works unchanged. See Quickstart.
  • Coding agent support: point OpenCode, Pi, Crush, or ForgeCode at Corti Models with the Corti CLI. See Use Corti Models in your coding agent.
Corti continuously upgrades the underlying foundation model behind each stable model ID, so you always get the best available model without changing your code. See How Corti models work.