JA4 q13d0310h3_55b375c5d22e_85bcee6495dd

q13d0310h3_55b375c5d22e_85bcee6495dd

Chrome — inferred from the User-Agent on its requests.

What this fingerprint encodes

Transport

QUIC

TLS version

TLS 1.3

Server name

server name sent

Cipher suites offered

3

Extensions offered

10

ALPN

h3

The hello it was computed from

Recovered because the proxy now stores the pre-hash JA3 string alongside the digest. Every number below came out of this client's ClientHello; anything we cannot name in the IANA registry is shown as its raw value rather than guessed at.

Version

TLS 1.2

Cipher suites 3

• TLS_AES_128_GCM_SHA256
• TLS_AES_256_GCM_SHA384
• TLS_CHACHA20_POLY1305_SHA256

Extensions 10

• supported_versions
• application_layer_protocol_negotiation
• signature_algorithms
• quic_transport_parameters
• supported_groups
• application_settings
• compress_certificate
• psk_key_exchange_modes
• key_share
• server_name

Named groups 3

• x25519
• secp256r1
• secp384r1

771,4865-4866-4867,43-16-13-57-10-17513-27-45-51-0,29-23-24,

Seen in live traffic

Connections

2

First seen

2026-08-24 14:10 UTC

Last seen

2026-08-24 14:11 UTC

Transport

QUIC

JA3 hashes absorbed

2

The same client, 2 different JA3s

This one JA4 covers 2 distinct JA3 hashes. A JA3 hashes the cipher and extension lists in the order they arrived, so a client that shuffles them — which Chrome and its derivatives do on purpose — produces a new JA3 almost every connection and fragments into what looks like 2 unrelated clients. JA4 sorts those lists before hashing, which is why all of it lands here instead.

• 2d7c1e83799c40b90cffd56424ed8662
• ee0e694b10a1bc263bb13b5b76e2bd53

User-Agents seen on this fingerprint

• Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/81.0.4044.138 Safari/537.36 21×
• Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) HeadlessChrome/114.0.5720.0 Safari/537.36 3×

A User-Agent is self-declared and trivially forged, so this names an observation rather than proving an identity. It is still the strongest signal available: the fingerprint comes off the TLS handshake and the User-Agent off the request that followed, and one client build keeping a stable JA4 while changing what it calls itself is a finding in its own right.

This entry is an observation, not a policy decision. It is here because the edge saw it, not because anyone reviewed it, and it blocks nothing on its own. Only the curated tier drives classification and banning.