YFarmX

AI Desk · Model Identity · Compare

Meta: Llama 4 Scout · Meta: Llama Guard 4 12B

Identity similarity: lowDivergent fingerprints · 30/50These models tokenise differently. Divergence on the high-information strings (CJK, emoji, digits) separates unrelated vocabularies within a handful of samples.

Meta: Llama 4 Scout recordMeta: Llama Guard 4 12B recordOpen in the interactive bench

Tokenizer fingerprint

30 of 50 mutually clean strings agree
Test stringMeta: Llama 4 ScoutMeta: Llama Guard 4 12BVerdict
en-prose1414match
en-long1716differs
spaces-2033match
spaces-6033match
tabs-2033match
newlines-2044match
mixed-ws66match
digits-934differs
digits-1245differs
digits-301011differs
digits-sep78differs
float-long910differs
zh-common1111match
zh-long1214differs
zh-rare1819differs
ja-kana99match
ja-kanji1211differs
ko109differs
ru1010match
ar1111match
he1010match
hi1212match
th88match
el1414match
emoji-basic1011differs
emoji-skin2020match
emoji-zwj-family1112differs
emoji-zwj-x33334differs
emoji-flags1617differs
emoji-prof1920differs
math2424match
boxdraw2122differs
combining1010match
cjk-ext-b1617differs
surrogates3737match
zalgo3131match
rtl-mix66match
py-code2525match
py-indent1515match
json2020match
html1617differs
regex4243differs
camel66match
snake66match
rare-word-x53030match
repeat-tok4040match
base643535match
hex1111match
url1717match
uuid2728differs

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
FieldMeta: Llama 4 ScoutMeta: Llama Guard 4 12BVerdict
Context length1310720163840× differs
Max output tokens1638416384✓ match
Supported parametersfrequency_penalty, logit_bias, max_tokens, min_p, presence_penalty, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_k, top_pfrequency_penalty, logit_bias, max_tokens, min_p, presence_penalty, repetition_penalty, response_format, seed, stop, temperature, top_k, top_p× differs
Defaults{}{}✓ match
Reasoning contract✓ match
Modalitiestext+image->texttext+image->text✓ match

Method: tokenizer fingerprinting, v1, 50 strings. This page is static and dated; the interactive bench compares any two of the 453 catalogue models.