毫秒API

api.holdai.top

毫秒API provides a stable, high-bandwidth API forwarding service for OpenAI-compatible models, including GPT, Claude, and Midjourney, with global server deployment and transparent pricing.

Models: 1184 models
From: $0.0015/M
Speed: 82 tok/s
Updated: 8/13/2026

Latency: 0.00 s
Created At: 8/13/2025

Features

DrawingTaskData Export

API Endpoints

Endpoint 1Historical / Unverified
https://api.gptbest.vip
Endpoint 2Historical / Unverified
https://hk-api.gptbest.vip
Endpoint 3Historical / Unverified
https://usage.gptbest.vip
Endpoint 4Historical / Unverified
https://api.holdai.top
Endpoint 5Historical / Unverified
https://api.bltcy.ai

Claim this provider

Verify ownership to unlock provider management features:

Edit provider name, content, and links
Get featured with priority traffic and visibility boost
Display a verified badge to build user trust

Leaderboard Rankings

Health

1.0%#245/717

Data as of Aug 13, 2026, 06:20 AMRankings are based on community-submitted tests and periodic health probes. Advisory only, not official data.

About 毫秒API

毫秒API is an API forwarding service that offers enterprise-grade, high-bandwidth access to AI models. It supports OpenAI-compatible interfaces, including GPT-3.5, GPT-4, Claude, Midjourney, and other models, with real-time updates. Key features include:

Global server deployment across 8 regions (e.g., US, Japan, Singapore) using CN2 lines and load balancing for low latency and high stability.
Transparent, usage-based pricing with a conversion rate of approximately 1.2 RMB to 1 USD, and VIP/SVIP discounts available.
Support for multiple model groups (e.g., reverse-engineered, Azure, pure OpenAI) to cater to different needs, with high concurrency handling up to 1 million TPM.
Full compatibility with OpenAI API protocols, enabling seamless integration into existing applications.
Additional services include Midjourney proxy support, 7*24-hour self-service recharge, and proxy distribution options.
Use cases include AI chat, file analysis, multimodal tasks, text-to-speech, and image generation via Midjourney.

Health Check

97%Recent availability

History (72 pts)

PastNow

API Benchmarks & Pricing

Compare 2 model rows across audit recency, latest speed tests, throughput, latency, and per-token pricing.

Model	Speed	Latency
chatgpt-4o-latest	100.7 t/s	3.47 s
gemini-2.0-pro-exp	70.1 t/s	7.13 s

Showing 2 of 2 model rows

Recent Test Records

Time	Model	Speed	Latency
Feb 20, 03:23 AM	gemini-2.0-pro-exp	70.15 tok/s	7.13s
Feb 20, 03:21 AM	chatgpt-4o-latest	100.72 tok/s	3.47s

Similar API Provider Alternatives to Compare

Compare 毫秒API alternatives against 6 nearby API providers using 3190 LMSpeed signals across shared model coverage, pricing, benchmark speed, uptime, and free-model availability.

Provider	Why compare	Models	Free	Avg price	Speed	30d uptime
毫秒API haomiao-api 毫秒API provides a stable, high-bandwidth API forwarding service for OpenAI-compatible models, including GPT, Claude, and Midjourney, with global server deployment and transparent pricing.	Current provider baseline	1,184	6	$0.0015/M	82 tok/s	9940%
FreeModel freemodel FreeModel is an OpenAI-compatible API gateway that provides access to frontier AI models through a unified endpoint, with free usage options for developers.	Faster measured speed Higher 30-day availability	17	0	N/A	253 tok/s	9980%
霁风的小圈 cpa-2006038-xyz A personal blog and project site for sharing technical projects, essays, and notes.	Faster measured speed Higher 30-day availability	11	0	N/A	102 tok/s	9970%
Deno Deploy Proxy deno-deploy-proxy An AI API proxy deployed on Deno Deploy, providing OpenAI-compatible model access.	Faster measured speed Higher 30-day availability	1	0	N/A	180 tok/s	9960%
Gitee AI gitee-ai Gitee AI provides a platform for AI model hosting, inference, and deployment with access to open-source LLMs.	Faster measured speed	260	0	N/A	88 tok/s	7270%
智谱 AI zhipu-ai Zhipu AI provides the GLM series of large language models including GLM-4, ChatGLM, and CodeGeeX for text, code, and multimodal tasks.	Higher 30-day availability	35	0	N/A	79 tok/s	9950%
Huawei Cloud huawei-modelarts Huawei ModelArts MaaS (Model as a Service) platform offering Pangu series models and third-party LLMs via OpenAI-compatible API.	Higher 30-day availability	14	0	N/A	31 tok/s	9970%

FAQ

Contact customer service

🐧: 1340650660 WeChat: llmh217 If you have any questions, don't be shy and ask directly. Our customer service will do their best to serve every customer.

What is quota? What is multiplier? How is the system calculated?

Quota = group multiplier * model multiplier * (prompt tokens + completion tokens * completion multiplier). The completion multiplier is fixed at 1.33 for GPT-3.5 and 2 for GPT-4, consistent with the official. In non-streaming mode, the official API returns the total tokens consumed, but note that prompt and completion have different multipliers. The multiplier is the billing method of this system, not a multiple of the official price. The price shown in this system is the final price, and billing is transparent.

About high concurrency

The system supports high concurrency by default. Feel free to go all out and make as many calls as you need.

Why is there no output or completion tokens after a request?

Possible reasons: 1. The client disconnected just before output started. 2. A tool call or function call occurred. 3. OpenAI directly returned [Done], usually due to policy or safety-related refusals. You need to check the finish_reason or content to determine the cause.

How can I tell if it is a GPT-4 model?

You can test with the following logical question: Question: What is the relationship between Lu Xun and Zhou Shuren? GPT-3.5: Lu Xun and Zhou Shuren are two different people. GPT-4: Lu Xun and Zhou Shuren are the same person.

What is the difference between the groups?

The groups are based on different channels, including official APIs and reverse-engineered ones. Official channels may also be divided into OpenAI official, Azure official, AWS official, etc. Reverse-engineered channels are the most cost-effective, while official channels are the most stable. You can choose according to your needs.

Notes

Health checks: Scope: the 72-hour chart and recent availability measure API connectivity only. Each bar summarizes one hour of checks. Targets: LMSpeed tries the configured health check URL and provider status URL first, then API endpoints derived from known API hosts and recent speed-test base URLs. A website host is considered only when it looks like an API endpoint. Probe steps: each candidate goes through DNS lookup, TCP connection, TLS handshake for HTTPS, and an HTTP HEAD request with redirects followed. Probing stops after the first reachable candidate. Reachable criteria: every required network step must succeed. An HTTP response below 500 is treated as reachable, including 401 because it confirms that an authenticated API endpoint responded, except for statuses classified as blocked. Blocked results: HTTP 403, 429, 521, 525, and 530, plus detected WAF or Cloudflare challenges, are shown as blocked and excluded from availability calculations because LMSpeed cannot determine whether the API itself is down. Model availability: when a dedicated test key is configured, LMSpeed sends an authenticated GET request to a derived /models endpoint and compares returned model IDs with this provider's listed models. These per-model results appear in Models & Pricing and are not included in the provider connectivity percentage. Timeouts: TCP connection, TLS handshake, HTTP connectivity, and model requests each use a 20-second timeout. A full run can take longer when several candidates are tried. Frequency: a background worker checks all providers every 5 minutes by default. The 72-hour chart combines those samples into hourly bars, and the schedule may be changed by the service operator. Limit: automated samples are not an SLA and do not guarantee account quota, every model, every region, or successful completion requests. Check the provider's own status page before making operational decisions.
Domain Rating data is sourced from Ahrefs. It is a 0–100 backlink-based domain strength signal and does not measure API speed or reliability.
Announcements and FAQ are read from this provider's NewAPI status snapshot when available. LMSpeed stores the original content and optional English translations from the provider status source, then shows the localized fields on this page.