Ship

A best execution endpoint for intelligence.

• Simple integration — prepend ship-like/ to a supported model name
• Save 50% on cost — our research-driven inference-time optimizations reduce your LLM spend
• Same capability and behavior — we carefully measure existing models and ensure our inference-time optimizations produce distributionally identical outputs
• Quality SLA — we're the only company in the LLM space that offers a quality SLA. If you're interested, contact our team at research@thesean.ai

Getting Started

If you're already using Thesean, enabling Ship is a one-line change: use a supported ship-like/ model name.

import openai

client = openai.OpenAI(
    base_url="https://api.thesean.ai/v1",
    api_key=THESEAN_API_KEY,
)

response = client.responses.create(
    model="ship-like/gpt-5.6-sol",
    input="Explain the Navier-Stokes equations.",
)

print(response.output_text)

No new SDKs. No configuration. No new endpoints. The API is identical to Thesean — you just get cheaper results at the same quality.

With deeper integration through the Feedback API, Ship learns what matters in your specific application and unlocks quality improvements, reliability gains, and latency reduction beyond what any single model can provide.

What Makes Us Different

Ship is an inference-time optimization tool. It treats models as frozen after training, then optimizes how they are called—for example, by introducing steering vectors into the models or performing prompt optimization.

These inference-time optimizations are powered by Thesean's research into mechanistic interpretability.

For more information, see:

• Ship launch blog post
• The Ship changelog