ChatGPT-User: User-Agent Signature, Runtime Mechanics & Verification

Q: Does ChatGPT-User execute JavaScript or hydrate client-side SPAs?

No. Unlike Google's Web Rendering Service, ChatGPT-User operates as a lightweight, low-latency fetcher. It parses raw server-rendered HTML. SPAs that require client hydration return blank containers to the crawler and forfeit citations.

AI Bot Profile · OpenAI Runtime

ChatGPT-User: User-Agent Signature, Runtime Telemetry & Verification

ChatGPT-User is OpenAI’s synchronous verification crawler. Triggered directly when a human user submits a search query in ChatGPT, this bot fetches live web pages to extract facts and generate footnote citations.

Live User-Inference Trigger · High Citation Value

Operator: OpenAI LLC

•

Observed Hits: 2,970 Sessions

•

Bot Traffic Share: 46.8% of Bot Volume

•

Network ASN: AS16509 / AS20473

1. Technical User-Agent Identification

ChatGPT-User identifies itself via a standardized user-agent string. When verifying requests at your edge router or CDN, inspect the incoming User-Agent header:

Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot

Verified Network Signatures & Official IP Feeds: In our production edge logs across 110 days, authentic ChatGPT-User traffic originated exclusively from OpenAI’s dedicated cloud infrastructure:

Official IP Ranges JSON:https://openai.com/chatgpt-user.json (maintained by OpenAI for firewall allowlists).
Autonomous System Numbers (ASNs):AS16509 (Amazon AWS) and AS20473 (The Constant Company / Vultr).
Reverse DNS Validation: All legitimate IP addresses resolve backwards to *.openai.com hostnames.
HTTP Protocol: Connects primarily over HTTP/2 with TLS 1.3 cipher suites.

OpenAI Crawler Disambiguation Matrix

OpenAI uses three distinct user-agent tokens for separate operational tasks. Managing them individually gives site owners precise control over their data:

User-Agent Token	Operational Purpose	Trigger Mechanism	Impact of Disallow
`ChatGPT-User`	Live search verification & source grounding	Active human prompt in ChatGPT	Blocks live citation attribution
`OAI-SearchBot`	SearchGPT web indexing & answer caching	Scheduled algorithmic web crawl	Removes pages from SearchGPT index
`GPTBot`	Foundational base model pre-training	Infrequent offline batch sweeps	Prevents training on your content

2. Zelitho Production Telemetry (N=13,971 Sessions)

Across our benchmark tracking window (June 11 to September 30, 2026), ChatGPT-User was the single most active bot hitting our server:

2,970

Verified Fetches

46.8% of total bot requests

2.93:1

Ratio vs. Googlebot

Outpaced Googlebot (1,012 hits)

2 to 4

Burst Fetches

Concurrent requests per prompt

99.1%

HTTP 200 OK

Low friction SSR ingestion

Burst Ingestion Dynamics: When an enterprise user asks ChatGPT a complex comparative question (e.g., “Compare software pricing and API latency”), the model fans out into sub-queries. We observed ChatGPT-User issuing 2 to 4 rapid parallel requests within 1,200ms to fetch distinct sections of the target site.

3. Ingestion & Chunking Mechanics

To earn top-3 footnote citations from ChatGPT-User, engineering teams must align their content with its parsing rules:

No Client-Side JavaScript Hydration:ChatGPT-User does not execute client-side JavaScript. If your content is rendered exclusively through React client components (useEffect) or Vue SPAs without Server-Side Rendering (SSR), the bot receives an empty container and immediately leaves.

256-Token Context Windowing: After extracting clean HTML, OpenAI’s retrieval pipeline feeds text into cross-encoder rerankers in ~256-token windows. When claims, data points, and technical mechanisms live in the same 150-word syntactic chunk, the reranker assigns maximum relevance and attributes the citation.

4. Configuration & Spoof Protection

To ensure ChatGPT-User can access your content for live search answers without opening your site to malicious scrapers, use these directives:

Recommended robots.txt Configuration

# Allow ChatGPT-User to fetch live verification and cite your content User-agent: ChatGPT-User Allow: / Disallow: /admin/ Disallow: /private/ Disallow: /api/

Cloudflare WAF Rule to Block Spoofed User-Agents

Malicious scrapers often spoof the ChatGPT-User string to bypass basic firewalls. Use this Cloudflare WAF rule to challenge or block unverified bots:

(http.user_agent contains "ChatGPT-User" and not cf.client.bot)

Action: Set to Managed Challenge or Block. Cloudflare’s verified bot system automatically checks reverse DNS to confirm legitimate OpenAI origins.

Frequently Asked Questions: ChatGPT-User

How can I verify authentic ChatGPT-User requests against spoofed scrapers?

Cross-reference client IP addresses against OpenAI’s official endpoint at https://openai.com/chatgpt-user.json, or perform reverse DNS lookups to confirm the PTR record resolves to a valid *.openai.com hostname.

Does ChatGPT-User execute JavaScript or hydrate client-side SPAs?

No. Unlike Google’s Web Rendering Service, ChatGPT-User is a lightweight, low-latency fetcher. It parses raw server-rendered HTML. SPAs that require client hydration return blank containers to the crawler and forfeit citations.

Will blocking ChatGPT-User hurt our Google Search rankings?

No. Googlebot and ChatGPT-User operate independently. Blocking ChatGPT-User in robots.txt has zero effect on Google PageRank or traditional organic search visibility, though it prevents your brand from being cited in ChatGPT Search.

Source Data: This telemetry was collected and analyzed as part of The 2026 Global AI Search & LLM Retrieval Benchmark (13,971 Production Edge Sessions Analyzed by Zelitho Research).

Audit Your Site for ChatGPT-User Readiness

Zelitho optimizes enterprise documentation and B2B websites for live LLM retrieval. Measure your Live Citation Velocity (LCV) and eliminate retrieval friction.

Explore Zelitho AI Growth Plans →