StepFun vs General Compute

StepFun builds its own multimodal models, many under Apache 2.0. General Compute serves other labs' open models with fast ASIC decode.

By The Subconscious Team · Updated October 27, 2026

StepFun vs General Compute: key differences

StepFun's Step 5 Preview is a 600B mixture-of-experts model with 1M context and image input, and Step 3.7 Flash costs $0.20 in and $1.15 out per million tokens with 256K context at about 128 tokens per second. First-party inference is hosted in China, with OpenRouter as another route. General Compute serves text models from MiniMax and DeepSeek plus GPT-OSS 120B in us-west-2, with GPT-OSS 120B at $0.21 in and $0.79 out. In its own benchmark it measured over 1,000 tokens per second on that model.

Pick StepFun for low-cost image and video understanding, context up to 1M, or Apache 2.0 weights to self-host. Pick General Compute for fast text output from a US region, private LoRA or fine-tuned checkpoints, or dedicated racks with a throughput SLA.

What StepFun and General Compute do

StepFun

StepFun is a Shanghai AI lab known for efficient multimodal models, with a mix of proprietary API models and open-weight releases. In September 2026 it launched Step 5 Preview, its new flagship for agentic work: a 600B-total, 27B-active mixture-of-experts model with 1M context and text and image input, which Artificial Analysis scores level with Kimi K3. Its cheaper 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 5 Preview, Step 3.7 Flash

Full StepFun profile

General Compute

General Compute is a San Francisco inference cloud built on alternative chips. It splits each request across hardware, with NVIDIA GPUs handling prefill and ASICs from Cerebras, SambaNova, Positron AI and d-Matrix handling decode. In its own open benchmark on GPT-OSS 120B it measured 1,025 to 1,234 output tokens per second against 100 to 121 on an unnamed GPU cloud, and up to 16.2x faster end to end. Founders Finn Puklowski and Jason Goodison raised a $15M seed led by FUSE, an up to $400M debt facility from Upper90 and, on October 27, 2026, a $100M Series A co-led by Blackbird and Abstract.

Example models: MiniMax M2.7, DeepSeek V3.2, GPT-OSS 120B

Full General Compute profile

Should you choose StepFun or General Compute?

StepFun

Choose StepFun for

Cheap multimodal models
Up to 1M context on Step 5 Preview
Apache 2.0 weights to self-host

General Compute

Choose General Compute for

Fast text decode in a US region
Hosting your own LoRA checkpoints
Throughput SLAs on dedicated racks

StepFun vs General Compute at a glance

Attribute	StepFun	General Compute
Model access	Open (Apache 2.0) and API models	Open weights
Flagship models	Step 5 Preview, Step 3.7 Flash	MiniMax M2.7, DeepSeek V3.2, GPT-OSS 120B
Speed	~128 tok/s on Step 3.7 Flash	1,000+ tok/s on GPT-OSS 120B (vendor)
Price	$0.20 in, $1.15 out (Step 3.7 Flash)	$0.21–$0.28 in, $0.38–$1.20 out per 1M
Customization	Open weights to fine-tune	Custom LoRA and fine-tuned checkpoints
Deployment	First-party API, OpenRouter	Serverless API, dedicated racks, bare metal
Long context	Up to 1M (Step 5 Preview); 256K on Step 3.7 Flash	192K (MiniMax M2.7)

Frequently asked questions

What is the difference between StepFun and General Compute?

StepFun builds its own multimodal models, many under Apache 2.0. General Compute serves other labs' open models with fast ASIC decode.

When should I choose StepFun over General Compute?

Cheap multimodal models; Up to 1M context on Step 5 Preview; Apache 2.0 weights to self-host.

When should I choose General Compute over StepFun?

Fast text decode in a US region; Hosting your own LoRA checkpoints; Throughput SLAs on dedicated racks.

Is StepFun or General Compute cheaper?

StepFun: $0.20 in, $1.15 out (Step 3.7 Flash). General Compute: $0.21–$0.28 in, $0.38–$1.20 out per 1M. The cheaper choice depends on the model and workload.

Which has more context, StepFun or General Compute?

StepFun: Up to 1M (Step 5 Preview); 256K on Step 3.7 Flash. General Compute: 192K (MiniMax M2.7).

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 StepFun Subconscious vs General Compute