Commercial Hub presents and compares third-party API / Gateway options. Partnership, promotion or Affiliate relationships are always labelled; information submitted by a Provider has been manually reviewed but is not official certification. A commercial relationship can buy visibility, but it can never buy Official Core benchmarks, rankings, test conclusions or the suppression of negative records. Prices and Offers are governed by the Provider's own website.Read the disclosure rules →

ENTERPRISE COMPUTE · FACT DIRECTORY

Enterprise compute & channel directory

Unranked factual channel labels (no tier numbers, no paid placement): official direct, institutional compute clouds, aggregator gateways, model asset libraries and vertical search are listed in separate tables; the enterprise side and the community-watch side never share a table.

All listed providers are independently verified for network connectivity and official enterprise contracting channels; links marked "Ad · Sponsored" are official partnerships and do not affect your price or service terms.

[Official Direct]

Official Direct

Official endpoints and product entries of licensed cloud vendors; links marked "Ad · Sponsored" are partnerships that do not affect your price.

6 entries

Alibaba Cloud Model Studio (Bailian)

Alibaba Cloud

New-user per-model free quota (typically 1M tokens per model), 90-day validity (per official terms)
Qwen family and mainstream open models, OpenAI-compatible API
Direct Alipay / bank-card billing with transparent invoices

Claim free new-user quota Verified 2026-10-15

Tencent Cloud

Tencent Hunyuan models via official endpoints
GPU and lightweight cloud servers with new-user pricing
Bank-card / enterprise billing with official SLA

Claim free new-user quota Verified 2026-10-14

Volcano Ark (Volcengine)

ByteDance · Volcengine

Official channel

Collaboration reward program: daily token rebate (up to 5M/day per model, per official terms)
Official endpoints for Doubao and mainstream models
Free trial quota for verified new users

Official entry Verified 2026-10-14

Zhipu BigModel

Zhipu AI

Official channel

GLM-4-Flash series is permanently free
New-user resource pack on the open platform (per official terms)
Direct China-mainland access, strong Chinese-language performance

Official entry Verified 2026-10-14

DeepSeek

Official channel

Official open platform with an OpenAI-compatible API
General chat and reasoning models with a deep-thinking mode (per official docs)
Usage-based billing with off-peak discounts (per official pricing page)

Official entry Verified 2026-10-15

Moonshot Kimi

Moonshot AI

Official channel

Official Kimi open platform with an OpenAI-compatible API
Long-context flagship models (model IDs and specs per official docs)
Usage-based billing in CNY; file-extraction APIs are free for a limited time (per official terms)

Official entry Verified 2026-10-15

[Institutional Compute Cloud]

Institutional Compute Cloud

9 entries

Provider	Service type	Hosted models · weight owner	Direct from CN	Payment	Verified	Official entry
Groq	Third-party inference cloud (custom LPU chips)	meta-llama/Llama-3.3-70B-Versatile · Owner: Meta deepseek-ai/DeepSeek-R1-Distill-Llama-70B · Owner: DeepSeek meta-llama/Llama-3.1-8B-Instant · Owner: Meta	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗
SiliconFlow	Domestic GPU inference cloud	Qwen/Qwen2.5-72B-Instruct · Owner: Alibaba Cloud (Qwen) deepseek-ai/DeepSeek-V3 · Owner: DeepSeek deepseek-ai/DeepSeek-R1 · Owner: DeepSeek	Reachable	Alipay, WeChat Pay	2026-10-15	Website ↗ Model docs ↗
Together AI	Open-weight model inference cloud	meta-llama/Llama-3.3-70B-Instruct-Turbo · Owner: Meta deepseek-ai/DeepSeek-R1 · Owner: DeepSeek meta-llama/Llama-3.1-8B-Instruct-Turbo · Owner: Meta	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗
Cerebras	Inference cloud (wafer-scale CS chips)	meta-llama/llama3.3-70b · Owner: Meta Qwen/qwen2.5-coder-32b · Owner: Alibaba Cloud (Qwen) meta-llama/llama3.1-8b · Owner: Meta	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗
Fireworks AI	Open-weight inference cloud (enterprise)	meta-llama/llama-v3p3-70b-instruct · Owner: Meta deepseek-ai/deepseek-r1 · Owner: DeepSeek meta-llama/llama-v3p1-8b-instruct · Owner: Meta	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗
Novita AI	GPU inference cloud (value tier)	meta-llama/llama-3.3-70b-instruct · Owner: Meta deepseek-ai/deepseek-r1 · Owner: DeepSeek deepseek-ai/deepseek-v3 · Owner: DeepSeek	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗
NVIDIA NIM	NVIDIA official inference microservices (NIM)	meta/llama-3.3-70b-instruct · Owner: Meta nvidia/llama-3.1-nemotron-70b-instruct · Owner: NVIDIA deepseek-ai/deepseek-r1 · Owner: DeepSeek	Intermittent	International credit card	2026-10-15	Website ↗ Model docs ↗
DeepInfra	Open-weight inference cloud (low price)	meta-llama/Llama-3.3-70B-Instruct · Owner: Meta deepseek-ai/DeepSeek-R1 · Owner: DeepSeek deepseek-ai/DeepSeek-V3 · Owner: DeepSeek	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗
SambaNova	Enterprise inference cloud (custom accelerators)	meta-llama/Llama-3.3-70B-Instruct · Owner: Meta deepseek-ai/DeepSeek-R1 · Owner: DeepSeek meta-llama/Llama-3.1-8B-Instruct · Owner: Meta	Intermittent	International credit card	2026-10-15	Website ↗ Model docs ↗

All models above are hosted (third-party) weights; the gold "latest flagship" badge convention does not apply to these entries.

[Aggregator Gateway]

Aggregator Gateway

1 entries

Provider	Service type	Hosted models · weight owner	Direct from CN	Payment	Verified	Official entry
Agnes AI	AI aggregator gateway	agnes-ai/agnes-3.0-pro · Owner: Agnes AI agnes-ai/agnes-2.5-flash · Owner: Agnes AI agnes-ai/agnes-image-2.1-flash · Owner: Agnes AI	Not reachable directly	Undisclosed	2026-10-15	Website ↗ Model docs ↗

All models above are hosted (third-party) weights; the gold "latest flagship" badge convention does not apply to these entries.

[Model Asset Library]

Model Asset Library

1 entries

Provider	Service type	Hosted models · weight owner	Direct from CN	Payment	Verified	Official entry
Hugging Face	Global model asset library & inference router	meta-llama/Llama-3.3-70B-Instruct · Owner: Meta Qwen/Qwen2.5-Coder-32B-Instruct · Owner: Alibaba Cloud (Qwen) deepseek-ai/DeepSeek-R1 · Owner: DeepSeek	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗

All models above are hosted (third-party) weights; the gold "latest flagship" badge convention does not apply to these entries.

[Vertical Search]

Vertical Search

1 entries

Provider	Service type	Hosted models · weight owner	Direct from CN	Payment	Verified	Official entry
Perplexity	Vertical search-augmented model lab	sonar-reasoning-pro sonar-pro sonar	Not reachable directly	International credit card	2026-10-15	Website ↗ Model docs ↗

All models above are hosted (third-party) weights; the gold "latest flagship" badge convention does not apply to these entries.

Open the community watch directory →Comparison discipline →

Factual public-information listing: zero affiliate deals, no top-up entry; status subject to each provider's official site.