QAX-GPT 奇安信安全大模型
Qi-Anxin's security LLM, sold as a SOC robot and agent platform in China
Overview
QAX-GPT is the security-specialised large language model of Chinese cybersecurity vendor Qi-Anxin, sold mainly as the QAX-GPT Security Robot for alert triage, investigation, task generation and reporting in security operations centres. It also underpins Qi-Anxin's newer agent and digital-employee products. 奇安信 QAX-GPT 安全大模型，面向政企安全运营，报价需咨询。
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 14 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
What you can produce with QAX-GPT 奇安信安全大模型
- Around-the-clock triage of security alerts, separating high-value threats from noise
- Automated investigation that queries linked Qi-Anxin products for context
- Executable response tasks such as blocking, patching and malware clean-up for operations staff
- Weekly and monthly security reports with ATT&CK and attacker-view sections
- Conversational question-answering about current security posture
- A base model and agent platform for Qi-Anxin's digital-employee and threat-analysis products
QAX-GPT is built for one job, clearing the alert backlog in a large security operations centre, and Qi-Anxin has kept extending it: from a triage robot in 2023 to an agent platform with MCP tools, digital-employee products and a 7B model small enough to run on-premises. The figures behind it, 60 times analyst speed and a top CyberSec-Eval ranking, come from Qi-Anxin itself.
For buyers outside mainland China it is hard to evaluate: no public API, no published price and Chinese-first materials. Worth a pilot for organisations already running Qi-Anxin products; otherwise a reference point rather than a purchase.
What we found
QAX-GPT is the security-specialised large language model built by Qi-Anxin (奇安信), one of China's largest cybersecurity vendors, and the product most buyers meet is the QAX-GPT Security Robot (QAX-GPT安全机器人系统). Qi-Anxin launched it as Q-GPT in August 2023, with BOE and Geely signing as the first customers, and put a new version on general sale across all industries in March 2024. Its product page describes a model of close to a hundred billion parameters, and in September 2026 Qi-Anxin said it had compressed the core model from 32B to 7B parameters for on-premises deployment while keeping 98.43% of its judgement quality.
The robot is aimed at alert fatigue in large security operations centres. It triages alerts around the clock, investigates by querying linked Qi-Anxin products such as Tianyan, Tianqing and its endpoint and server tools, turns findings into concrete tasks for operations staff, and drafts weekly and monthly reports. Qi-Anxin's March 2024 figures are the ones usually quoted: alert judgement more than 60 times faster than an analyst, a false-positive rate about half a human's and a miss rate around 5% of one. Those are vendor numbers. So is the claim on its English site that a QGPT model ranked first overall on the CyberSec-Eval leaderboard with eight category wins. In April 2024 it received one of the first large-model security assessment certificates from a Chinese industry association and the China Software Testing Center.
Since then the model has become a base layer. Qi-Anxin now sells the robot as an agent platform with retrieval, workflow orchestration and MCP tools, plus newer products built on it, including a security-operations digital employee and a threat-analysis digital expert offered as SaaS.
Nothing is self-serve. Prices are quoted through Qi-Anxin sales after a consultation or trial request, materials are Chinese-first, and there is no public API for outside developers. It is relevant mainly to organisations in mainland China already running Qi-Anxin products.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
About this listing
This entry was compiled from publicly available data including QAX-GPT 奇安信安全大模型's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with QAX-GPT 奇安信安全大模型 unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to QAX-GPT 奇安信安全大模型 directly →
Spotted something out of date? Suggest an update →
Alternatives to QAX-GPT 奇安信安全大模型
More in Security & Safety
