Ottili AI

The Ottili Cairn model

Ottili Cairn is the premium price-performance model of Ottili AI: stronger reasoning and agentic capability for demanding software and analysis work. Status, use cases, routing through Ottili Auto, comparison with Vale and Spire, and billing via AI credits.

Overview

Ottili Cairn* is the premium price-performance tier of the Ottili AI Platform. It is built for demanding software and analysis work — strong reasoning, agentic coding, complex refactors and multi-step reasoning. Within the public Ottili AI model family, Cairn plays the role of the premium tier.

Cairn is generally available (GA)* — publicly documented with the maturity status Live*. It is part of a clearly tiered model lineup:

Ottili Vale* — efficient, everyday, cost-effective (Live/GA).
Ottili Cairn* — the premium tier for demanding engineering and analysis work (Live/GA).
Ottili Auto* — the adaptive router that picks Vale or Cairn per request (Live/GA).
Ottili Spire* — the reserved flagship model; currently reserved, not yet available*.

Cairn is the model Ottili Auto selects for heavy, agentic and reasoning-heavy work, and is the default for Ottili Coder Build Mode*, Review Mode* and MEE Mode*. This article explains what Cairn is, where it is used, how it is routed through Ottili Auto, how it compares across the lineup, and how billing works through AI credits.

What Ottili Cairn is

Ottili Cairn (current public version: Ottili Cairn 1.2*) is the premium tier of the Ottili AI Platform. The public tagline reads: *"Premium price-performance tier for demanding software and analysis work — stronger reasoning and agentic capability."*

Cairn is built for work that asks for more than everyday fixes:

Large repository analysis
Architecture planning
Multi-step coding and complex refactors
Deep reasoning and analysis
Production readiness and security reviews

Cairn is not* the efficient tier (that is Ottili Vale, behind deepseek-v4-flash) and not* the flagship (that is the reserved Ottili Spire, which has no public backing model and is never GLM-5.2). Its single differentiator versus Vale is the reasoning capability (supports_reasoning = True), carried on the backing model deepseek-v4-pro.

Status and availability

Cairn carries the maturity status Live (General Availability)*. That means it is generally available, fully supported, and intended for real workflows.

Availability by plan:

Free*: Cairn is not* included on the Free plan — only the efficient Vale is available there.
Starter*: Vale plus Cairn.
Pro, Business, Enterprise*: Vale plus Cairn (and Spire once the flagship is launched — currently reserved).

Maturity is expressed through Ottili ONE's canonical vocabulary: Concept, Planned, In Development, Private Beta, Beta, Rolling Out, Live, Deprecated, Disabled. Cairn is Live*; Spire is reserved* (not yet available). How Ottili ONE labels maturity publicly is described in [Understand feature status labels](/docs/understand-feature-status-labels).

Technical specifications

Public model registry (stable for the current version Ottili Cairn 1.2*):

Canonical public id: ottili/cairn-1.2 (stable — clients must pin this)
Stable family alias: ottili/cairn → ottili/cairn-1.2 (fail-closed alias resolution)
Tier: premium
Backing model: deepseek-v4-pro via Ottili-owned* DeepSeek routing
Status: active (general_availability)
Context window: 1,000,000 tokens
Max output tokens: 384,000 tokens
Capabilities: chat, streaming, tools, JSON mode, reasoning

Cairn compared

Model	Tier	Status	Role
Ottili Vale 1.2	efficient	Live / GA	everyday, baseline, cost-effective
Ottili Cairn 1.2	premium	Live / GA	demanding engineering and analysis work
Ottili Auto	automatic	Live / GA	picks Vale or Cairn per request
Ottili Spire 1.2	flagship	reserved	not yet available

Cairn is the premium tier for serious software and analysis work. Auto is not a model of its own but a router: it decides per request. Spire is reserved and deliberately has no public backing model.

Routing: Ottili Auto

When no specific model is chosen, Ottili Auto* takes over selection. The adaptive router distributes requests by task type:

Everyday fixes → Vale
Heavy tasks → Cairn
Agentic / multi-step → Cairn
Reasoning tasks → Cairn

Result: everyday work runs cost-effectively on Vale, while demanding work is routed automatically to Cairn. Cairn is also the default for Ottili Coder Build Mode*, Review Mode* and MEE Mode*.

Cost and credits

Cairn is the premium tier among the publicly available models. The price values are versioned configuration references* (not published retail prices), billed through the AI credit system:

Reference input price: 0.65 USD per 1 million tokens
Reference output price: 1.30 USD per 1 million tokens
Pricing version: 2026.07.vale-cairn-1

These values are configuration references only, for margin math and billing configuration. A final retail price is a business decision and is never presented as an immutable "fixed price". The exact, currently valid values are published in the public model registry of the Ottili AI Platform (ai.ottili.one) and billed through the AI credit system.

Billing works like this:

Prepaid via company credits*: Ottili AI is prefunded through a shared, company-scoped credit wallet.
Reserved and deducted per request*: the cost of a request is reserved and deducted from the wallet on execution; unused reservations are released.
Cost control*: per API key and per company, model limits, budget limits and rate limits are enforced.

How credits and plans relate is described in [Understand plans and credits](/docs/understand-plans-and-credits).

Security and data handling

Every Cairn workload stays within a single company boundary (company_id):

Company-scoped isolation* — Cairn is available to any company holding a valid AI key; an operator-applied restriction is honored fail-closed.
No prompt logging* — prompt and completion content is never logged; provider training opt-out is enforced.
Identity fidelity* — Cairn's system prompt always identifies itself as Ottili Cairn and never leaks the upstream provider.
Fail-closed routing* — Cairn has fallback_eligible = False by default; without a configured, budgeted paid fallback a request is reported blocked, never silently downgraded to a paid provider.
Audit* — every action is traceable.