# Subconscious vs Sail Research

> Sail and Subconscious both target long-horizon agents. Sail gets cheaper the longer you let each turn wait. Subconscious gets cheaper by processing fewer tokens, with fast turns.

Canonical: https://www.subconscious.dev/compare/subconscious-vs-sail-research · By The Subconscious Team · Updated October 11, 2026

## How they compare

This is the closest matchup in the guide, since both companies build for agents that run for hours. They take opposite routes. Sail sells throughput over latency. Customers pick a completion window, from the default low-latency asap window at 30 to 70% below traditional providers to off-peak flex at 55 to 85% below, and Sail packs as much work into each GPU as it can. Subconscious keeps the turn fast and shrinks the work instead. It prunes the KV cache and preserves suffix state, delivers 2x faster task completion and 50% to 80% lower cost than standard inference, and bills only the processed tokens. Sail trades time for price, while Subconscious cuts the tokens themselves.

The deciding question is whether anyone waits on the agent. For background work with no human in the loop, like a code scan that runs three to four hours, evals or offline research, Sail's windows and its Sailboxes for persistent compute are a strong fit, and its catalog of Kimi K2.6, GLM-5.3, GPT-OSS 120B, Qwen 3.6 and Gemma 4 is wider. Sail's slower, cheaper windows do not suit voice, live chat or interactive UIs. Subconscious covers that side: coding agents inside Claude Code, Cursor or GitHub Copilot and user-facing agent products, where a five-minute turn would not work.

## What each one does

### Subconscious

Subconscious is an MIT CSAIL spinout in Kendall Square that builds inference for long-horizon agents, the workloads where a single trace runs past 200K tokens and often into the millions. Its runtime drops in as a replacement for vLLM or SGLang. Instead of rereading an ever-growing context on every step, it prunes the KV cache and preserves suffix state, and Subconscious co-designs the runtime with post-trained model variants it calls Marathon. Against open models on standard inference, Subconscious delivers 2x faster task completion, delivers a 5M+ effective context window, cuts cost 50% and up to 80%, and scores neutral to 10% better on agentic benchmarks.

### Sail Research

Sail Research sells throughput over latency. Founders Neil Movva and Samir Menon built a serving stack that packs as much work as possible into every GPU, and customers state how long they can wait through completion windows. Its docs quote each window against traditional providers: the default asap window serves at low latency for 30 to 70% less, balanced for 45 to 80% less, and the off-peak flex window for 55 to 85% less. Batch requests default to flex, and LoRA and custom-tool requests default to balanced.

## Which is best, and when

### Choose Subconscious for

- Interactive coding agents where a minutes-long turn is too slow
- User-facing agent products with long contexts
- Cutting cost by processing fewer tokens, not by waiting for cheaper GPU time

### Choose Sail Research for

- Background agents that run for hours unattended
- Evals and offline research at 55 to 85% off
- Persistent agent compute through Sailboxes

## At a glance

| Attribute | Subconscious | Sail Research |
|---|---|---|
| Model access | Open weights | Open weights |
| Flagship models | GLM 5.3, DeepSeek V4.1 Flash | Kimi K2.6, GLM-5.3, GPT-OSS 120B |
| Speed | 2x faster task completion | Low-latency default; slower windows cheaper |
| Price | 50–80% lower cost; billed on processed tokens | 30–85% below other hosts by completion window |
| Customization | Marathon post-trained variants | Customer LoRA fine-tunes |
| Deployment | Managed API, dedicated, on-prem | API plus Sailboxes |
| Long context | 5M+ effective context | Varies by model |

## FAQ

### What is the difference between Subconscious and Sail Research?

Sail and Subconscious both target long-horizon agents. Sail gets cheaper the longer you let each turn wait. Subconscious gets cheaper by processing fewer tokens, with fast turns.

### When should I choose Subconscious over Sail Research?

Interactive coding agents where a minutes-long turn is too slow; User-facing agent products with long contexts; Cutting cost by processing fewer tokens, not by waiting for cheaper GPU time.

### When should I choose Sail Research over Subconscious?

Background agents that run for hours unattended; Evals and offline research at 55 to 85% off; Persistent agent compute through Sailboxes.

### Is Subconscious or Sail Research cheaper?

Subconscious: 50–80% lower cost; billed on processed tokens. Sail Research: 30–85% below other hosts by completion window. The cheaper choice depends on the model and workload.

### Which has more context, Subconscious or Sail Research?

Subconscious: 5M+ effective context. Sail Research: Varies by model.

## Try Subconscious

Subconscious speaks the OpenAI and Anthropic API formats. Base URL: https://api.subconscious.dev/v1. Docs: https://docs.subconscious.dev. Get an API key: https://platform.subconscious.dev/signin. Agent guide: https://www.subconscious.dev/agents.md.

Full profiles: [Subconscious](https://www.subconscious.dev/providers/subconscious.md), [Sail Research](https://www.subconscious.dev/providers/sail-research.md).