AI Desk · Model Identity · Compare

Ling-3.0-flash · Poolside: Laguna XS 2.1

Identity similarity: lowDivergent fingerprints · 23/50These models tokenise differently. Divergence on the high-information strings (CJK, emoji, digits) separates unrelated vocabularies within a handful of samples.

Ling-3.0-flash recordPoolside: Laguna XS 2.1 recordOpen in the interactive bench

Tokenizer fingerprint

23 of 50 mutually clean strings agree
Test stringLing-3.0-flashPoolside: Laguna XS 2.1Verdict
en-prose1414match
en-long1617differs
spaces-2033match
spaces-6033match
tabs-2053differs
newlines-20224differs
mixed-ws97differs
digits-999match
digits-121212match
digits-303030match
digits-sep1212match
float-long2222match
zh-common912differs
zh-long1020differs
zh-rare1820differs
ja-kana1014differs
ja-kanji1315differs
ko1213differs
ru1515match
ar1423differs
he1922differs
hi2526differs
th1525differs
el2124differs
emoji-basic1010match
emoji-skin2020match
emoji-zwj-family1111match
emoji-zwj-x33333match
emoji-flags1610differs
emoji-prof2020match
math2824differs
boxdraw3027differs
combining512differs
cjk-ext-b1616match
surrogates4037differs
zalgo3636match
rtl-mix911differs
py-code2626match
py-indent1818match
json2525match
html1616match
regex4542differs
camel56differs
snake66match
rare-word-x53136differs
repeat-tok4122differs
base643634differs
hex1816differs
url1919match
uuid3636match

Values are the marginal prompt-token cost of each string with the model's fixed overhead subtracted. An x marks a row excluded as corrupt for that model.

API surface

The declared serving contracts differ
FieldLing-3.0-flashPoolside: Laguna XS 2.1Verdict
Context length262144262144✓ match
Max output tokens3276832768✓ match
Supported parametersfrequency_penalty, include_reasoning, logit_bias, logprobs, max_tokens, min_p, presence_penalty, reasoning, repetition_penalty, response_format, seed, stop, temperature, tool_choice, tools, top_k, top_logprobs, top_pinclude_reasoning, max_tokens, reasoning, temperature, tool_choice, tools× differs
Defaults{}{}✓ match
Reasoning contract{"mandatory":false,"default_enabled":true}{"mandatory":false,"default_enabled":true}✓ match
Modalitiestext->texttext->text✓ match

Method: tokenizer fingerprinting, v1, 50 strings. Raw data on the data page. This page is static and dated; the interactive bench compares any two of the 422 catalogue models.