AI Desk · Model Identity · Compare
DeepSeek: DeepSeek V4 Pro 0813 · WizardLM-2 8x22B
Identity similarity: lowDivergent fingerprints · 32/50These models tokenise differently. Divergence on the high-information strings (CJK, emoji, digits) separates unrelated vocabularies within a handful of samples.
DeepSeek: DeepSeek V4 Pro 0813 recordWizardLM-2 8x22B recordOpen in the interactive bench
Tokenizer fingerprint
32 of 50 mutually clean strings agree| Test string | DeepSeek: DeepSeek V4 Pro 0813 | WizardLM-2 8x22B | Verdict |
|---|---|---|---|
| en-prose | 14 | 14 | match |
| en-long | 16 | 15 | differs |
| spaces-20 | 3 | 3 | match |
| spaces-60 | 3 | 3 | match |
| tabs-20 | 4 | 4 | match |
| newlines-20 | 4 | 4 | match |
| mixed-ws | 7 | 7 | match |
| digits-9 | 3 | 4 | differs |
| digits-12 | 4 | 5 | differs |
| digits-30 | 10 | 11 | differs |
| digits-sep | 7 | 8 | differs |
| float-long | 9 | 10 | differs |
| zh-common | 9 | 10 | differs |
| zh-long | 10 | 11 | differs |
| zh-rare | 18 | 19 | differs |
| ja-kana | 9 | 10 | differs |
| ja-kanji | 11 | 12 | differs |
| ko | 13 | 12 | differs |
| ru | 13 | 12 | differs |
| ar | 10 | 10 | match |
| he | 14 | 13 | differs |
| hi | 18 | 18 | match |
| th | 11 | 11 | match |
| el | 16 | 16 | match |
| emoji-basic | 10 | 10 | match |
| emoji-skin | 14 | 14 | match |
| emoji-zwj-family | 11 | 11 | match |
| emoji-zwj-x3 | 33 | 33 | match |
| emoji-flags | 16 | 16 | match |
| emoji-prof | 19 | 19 | match |
| math | 22 | 22 | match |
| boxdraw | 27 | 27 | match |
| combining | 10 | 10 | match |
| cjk-ext-b | 16 | 16 | match |
| surrogates | 33 | 33 | match |
| zalgo | 36 | 36 | match |
| rtl-mix | 6 | 6 | match |
| py-code | 25 | 25 | match |
| py-indent | 15 | 15 | match |
| json | 20 | 20 | match |
| html | 16 | 17 | differs |
| regex | 47 | 48 | differs |
| camel | 7 | 6 | differs |
| snake | 8 | 8 | match |
| rare-word-x5 | 26 | 26 | match |
| repeat-tok | 41 | 41 | match |
| base64 | 34 | 34 | match |
| hex | 12 | 12 | match |
| url | 17 | 17 | match |
| uuid | 27 | 28 | differs |
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| Field | DeepSeek: DeepSeek V4 Pro 0813 | WizardLM-2 8x22B | Verdict |
|---|---|---|---|
| Context length | 1048576 | 65535 | × differs |
| Max output tokens | 393216 | 8000 | × differs |
| Supported parameters | frequency_penalty, include_reasoning, logit_bias, logprobs, max_tokens, min_p, presence_penalty, reasoning, reasoning_effort, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_k, top_logprobs, top_p | frequency_penalty, max_tokens, presence_penalty, repetition_penalty, response_format, seed, stop, temperature, top_k, top_p | × differs |
| Defaults | {"temperature":1,"top_p":1} | {} | × differs |
| Reasoning contract | {"mandatory":false,"supported_efforts":["max","high","low"],"default_effort":"high"} | – | × differs |
| Modalities | text->text | text->text | ✓ match |
Method: tokenizer fingerprinting, v1, 50 strings. This page is static and dated; the interactive bench compares any two of the 486 catalogue models.