WizardLM-2 8x22B

microsoft/wizardlm-2-8x22b · MicrosoftNot yet established

Declared family: Mistral
Declared tokenizer tag: Mistral
Entered catalogue: 16 April 2024
Last tested: 22 August 2026
Measured signature: tk_a49eb75d

Compare with DeepSeek: DeepSeek V4 Pro 0813

Identity statement

WizardLM-2 8x22B has a clean measured fingerprint that currently sits alone: no other measured model shares its signature.

What we checked

Four different kinds of evidence, and what each one showed

We checked	What we found	Where it came from
How it counts tokens	Counted cleanly, and no other model in the catalogue counts the same way	tk_a49eb75d
How its API is set up	A combination of settings no other listing in the catalogue uses	api_1c686178
How much it can read and write	Reads up to 65,535 tokens at once, writes up to 8,000	its own listing
Who served our requests	Novita	we measured it, 22 August 2026

Closest measured models

Agreement across the mutually clean test strings

DeepSeek: DeepSeek V4 Pro 0813deepseek/deepseek-v4-pro-081332/50 · 64%DeepSeek: DeepSeek V4 Flash Vision Expdeepseek/deepseek-v4-flash-vision-exp32/50 · 64%DeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-073132/50 · 64%DeepSeek: DeepSeek V4 Flash 0423deepseek/deepseek-v4-flash32/50 · 64%StepFun: Step 3.5 Flashstepfun/step-3.5-flash32/50 · 64%StepFun: Step 3.7 Flashstepfun/step-3.7-flash32/50 · 64%

Click a row to open the full comparison.

The measured fingerprint

What each test string cost this model, in tokens

We sent WizardLM-2 8x22B fifty short pieces of text and recorded what each one cost it in tokens. The bars below are those costs. Two models built on the same tokenizer produce the same bars; a model built on a different one produces a different set, which is what makes this a fingerprint.

All 50 rows

English and whitespace

en-prose14

en-long15

spaces-203

spaces-603

tabs-204

newlines-204

mixed-ws7

Digits

digits-94

digits-125

digits-3011

digits-sep8

float-long10

CJK

zh-common10

zh-long11

zh-rare19

ja-kana10

ja-kanji12

ko12

Other scripts

ru12

ar10

he13

hi18

th11

el16

Emoji

emoji-basic10

emoji-skin14

emoji-zwj-family11

emoji-zwj-x333

emoji-flags16

emoji-prof19

Rare Unicode

math22

boxdraw27

combining10

cjk-ext-b16

surrogates33

zalgo36

rtl-mix6

Code

py-code25

py-indent15

json20

html17

regex48

camel6

snake8

Repetition and encodings

rare-word-x526

repeat-tok41

base6434

hex12

url17

uuid28

Overhead subtracted: 34 prompt tokens.

Declared record

What the catalogue claims about this model

WizardLM-2 8x22B is Microsoft AI's most advanced Wizard model. It demonstrates highly competitive performance compared to leading proprietary models, and it consistently outperforms all existing state-of-the-art opensource models. It is...

Modalities: text->text
Declared tokenizer: Mistral
Prompt price: $0.62 / M tokens
Completion price: $0.62 / M tokens

Supported parameters

frequency_penaltymax_tokenspresence_penaltyrepetition_penaltyresponse_formatseedstoptemperaturetop_ktop_p

defaults: {}

Catalogue entry Weights on Hugging Face

History

Every observation, kept as taken

16 April 2024Enters the OpenRouter catalogue declared as Mistral.
22 August 2026Fingerprinted in the YFarmX catalogue sweep · 50 of 50 strings measured clean.