Sail Research vs StepFun

Sail Research discounts open-model inference by completion window. StepFun is a Shanghai lab selling its own cheap multimodal Step models. Host with a price lever versus model maker.

By The Subconscious Team · Updated September 30, 2026

Sail Research vs StepFun: key differences

StepFun makes models; Sail Research serves other people's. StepFun's Step 3.7 Flash is a 198B mixture-of-experts vision-language model with 11B active, 256K context and an Apache 2.0 license, priced at $0.20 in and $1.15 out on StepFun's own API. Sail's catalog is a set of open text models, including Kimi K2.6, GLM-5, GPT-OSS 120B, Qwen 3.6 and Gemma 4, plus customer LoRA fine-tunes, and its lever is time. Customers who wait longer, up to an off-peak flex window, pay 30 to 80% less than the asap price.

Pick by workload shape and by where the traffic may go. StepFun is the stronger choice for image and video understanding on a budget, or for a small-active-parameter model to self-host. Its first-party inference is hosted in China, which rules it out for some buyers, and Western support is thin. Sail is built around long async agents, with Sailboxes for persistent compute and OpenAI and Anthropic-compatible APIs. It is a bad fit for any interactive UI, while StepFun's API has no such built-in wait.

What Sail Research and StepFun do

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. The priority window targets about a one-minute turn for roughly 30 to 50% off the immediate asap price. The default standard window targets about five minutes for 45 to 65% off. The flex window runs off-peak for 60 to 80% off.

Example models: Kimi K2.6, GLM-5

Full Sail Research profile

StepFun

StepFun is a Shanghai AI lab known for efficient multimodal models, with a mix of proprietary API models and open-weight releases. Its current workhorse, Step 3.7 Flash, came out in May 2026 as a 198B mixture-of-experts vision-language model with only 11B active parameters. It has 256K context, selectable reasoning levels, tool use and structured outputs, and it ships under Apache 2.0. StepFun's own API prices it at $0.20 in and $1.15 out per million tokens, and OpenRouter carries it too.

Example models: Step 3.7 Flash, Step3

Full StepFun profile

Should you choose Sail Research or StepFun?

Sail Research

Choose Sail Research for

Background agents that trade minutes of delay for deep discounts.
Batch evals on popular open text models.
Serving customer LoRA fine-tunes cheaply.

StepFun

Choose StepFun for

Low-cost vision and video understanding in agents.
Self-hosting Apache 2.0 weights with few active parameters.
Multimodal work needing 256K context on a real-time API.

Sail Research vs StepFun at a glance

Attribute	Sail Research	StepFun
Model access	Open weights	Open (Apache 2.0) and API models
Flagship models	Kimi K2.6, GLM-5, GPT-OSS 120B	Step 3.7 Flash, Step3
Speed	Minutes per turn by design	~128 tok/s on Step 3.7 Flash
Price	30–80% off by completion window	$0.20 in, $1.15 out (Step 3.7 Flash)
Customization	Customer LoRA fine-tunes	Open weights to fine-tune
Deployment	API plus Sailboxes	First-party API, OpenRouter
Long context	Varies by model	256K

Frequently asked questions

What is the difference between Sail Research and StepFun?

Sail Research discounts open-model inference by completion window. StepFun is a Shanghai lab selling its own cheap multimodal Step models. Host with a price lever versus model maker.

When should I choose Sail Research over StepFun?

Background agents that trade minutes of delay for deep discounts; Batch evals on popular open text models; Serving customer LoRA fine-tunes cheaply.

When should I choose StepFun over Sail Research?

Low-cost vision and video understanding in agents; Self-hosting Apache 2.0 weights with few active parameters; Multimodal work needing 256K context on a real-time API.

Is Sail Research or StepFun cheaper?

Sail Research: 30–80% off by completion window. StepFun: $0.20 in, $1.15 out (Step 3.7 Flash). The cheaper choice depends on the model and workload.

Which has more context, Sail Research or StepFun?

Sail Research: Varies by model. StepFun: 256K.

Related comparisons

Running long-horizon agents?

If your agents run past 200K tokens, compare both against Subconscious. Our inference stack treats a long-horizon trace as the primary workload, so speed, cost, and accuracy hold up deep into the trace.

Subconscious vs Sail Research Subconscious vs StepFun