AI Desk · Model Identity · Compare
Amazon: Nova 2 Lite · SpaceXAI: Grok Build 0.1
Identity similarity: lowDivergent fingerprints · 31/50These models tokenise differently. Divergence on the high-information strings (CJK, emoji, digits) separates unrelated vocabularies within a handful of samples.
Amazon: Nova 2 Lite recordSpaceXAI: Grok Build 0.1 recordOpen in the interactive bench
Tokenizer fingerprint
31 of 50 mutually clean strings agree| Test string | Amazon: Nova 2 Lite | SpaceXAI: Grok Build 0.1 | Verdict |
|---|---|---|---|
| en-prose | 14 | 14 | match |
| en-long | 15 | 16 | differs |
| spaces-20 | 3 | 3 | match |
| spaces-60 | 3 | 3 | match |
| tabs-20 | 3 | 3 | match |
| newlines-20 | 4 | 4 | match |
| mixed-ws | 6 | 6 | match |
| digits-9 | 3 | 3 | match |
| digits-12 | 4 | 4 | match |
| digits-30 | 10 | 10 | match |
| digits-sep | 7 | 7 | match |
| float-long | 9 | 9 | match |
| zh-common | 12 | 9 | differs |
| zh-long | 16 | 13 | differs |
| zh-rare | 18 | 18 | match |
| ja-kana | 8 | 9 | differs |
| ja-kanji | 10 | 10 | match |
| ko | 12 | 11 | differs |
| ru | 11 | 11 | match |
| ar | 10 | 16 | differs |
| he | 10 | 14 | differs |
| hi | 11 | 13 | differs |
| th | 9 | 13 | differs |
| el | 14 | 15 | differs |
| emoji-basic | 10 | 10 | match |
| emoji-skin | 11 | 14 | differs |
| emoji-zwj-family | 11 | 11 | match |
| emoji-zwj-x3 | 33 | 33 | match |
| emoji-flags | 14 | 16 | differs |
| emoji-prof | 18 | 20 | differs |
| math | 25 | 27 | differs |
| boxdraw | 24 | 24 | match |
| combining | 10 | 10 | match |
| cjk-ext-b | 16 | 16 | match |
| surrogates | 32 | 40 | differs |
| zalgo | 34 | 35 | differs |
| rtl-mix | 6 | 6 | match |
| py-code | 25 | 25 | match |
| py-indent | 15 | 15 | match |
| json | 20 | 20 | match |
| html | 16 | 16 | match |
| regex | 42 | 42 | match |
| camel | 6 | 6 | match |
| snake | 6 | 6 | match |
| rare-word-x5 | 30 | 26 | differs |
| repeat-tok | 21 | 22 | differs |
| base64 | 31 | 33 | differs |
| hex | 10 | 10 | match |
| url | 17 | 17 | match |
| uuid | 27 | 27 | match |
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 | Amazon: Nova 2 Lite | SpaceXAI: Grok Build 0.1 | Verdict |
|---|---|---|---|
| Context length | 1000000 | 256000 | × differs |
| Max output tokens | 65535 | 230400 | × differs |
| Supported parameters | include_reasoning, max_tokens, reasoning, stop, temperature, tool_choice, tools, top_k, top_p | include_reasoning, logprobs, max_tokens, reasoning, response_format, seed, structured_outputs, temperature, tool_choice, tools, top_logprobs, top_p | × differs |
| Defaults | {"temperature":null,"top_p":null,"frequency_penalty":null} | {"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null} | × differs |
| Reasoning contract | {"mandatory":false} | {"mandatory":true} | × differs |
| Modalities | text+image+file+video->text | text+image+file->text | × differs |
Method: tokenizer fingerprinting, v1, 50 strings. This page is static and dated; the interactive bench compares any two of the 486 catalogue models.