AI Desk · AI Fingerprinting · Compare
Inception: Mercury 2.5 · Inception: Mercury 2.5 Preview
Insufficient evidenceNot comparable yetAt least one of these records is declared unmeasurable (Inception: Mercury 2.5: unmeasurable:caching; Inception: Mercury 2.5 Preview: unmeasurable:caching), so no similarity verdict is published; the declared contracts below can still be compared.
Inception: Mercury 2.5 recordInception: Mercury 2.5 Preview recordOpen in the interactive bench
Tokenizer fingerprint
18 of 25 mutually clean strings agree| Test string | Inception: Mercury 2.5 | Inception: Mercury 2.5 Preview | Verdict |
|---|---|---|---|
| en-prose | x | x | – |
| en-long | 2 | 2 | match |
| spaces-20 | x | x | – |
| spaces-60 | x | x | – |
| tabs-20 | x | x | – |
| newlines-20 | x | x | – |
| mixed-ws | x | x | – |
| digits-9 | x | x | – |
| digits-12 | x | x | – |
| digits-30 | 15 | 16 | differs |
| digits-sep | x | x | – |
| float-long | 8 | 8 | match |
| zh-common | x | x | – |
| zh-long | x | x | – |
| zh-rare | 4 | 4 | match |
| ja-kana | x | x | – |
| ja-kanji | x | x | – |
| ko | x | x | – |
| ru | x | x | – |
| ar | x | x | – |
| he | x | x | – |
| hi | 6 | 6 | match |
| th | x | x | – |
| el | x | x | – |
| emoji-basic | x | x | – |
| emoji-skin | 15 | 16 | differs |
| emoji-zwj-family | 4 | 4 | match |
| emoji-zwj-x3 | 39 | 39 | match |
| emoji-flags | 10 | 10 | match |
| emoji-prof | 15 | 14 | differs |
| math | 15 | 15 | match |
| boxdraw | 14 | 14 | match |
| combining | x | x | – |
| cjk-ext-b | x | x | – |
| surrogates | 25 | 26 | differs |
| zalgo | 21 | 22 | differs |
| rtl-mix | x | x | – |
| py-code | 11 | 11 | match |
| py-indent | 4 | 4 | match |
| json | 11 | 11 | match |
| html | 2 | 2 | match |
| regex | 28 | 29 | differs |
| camel | x | x | – |
| snake | x | x | – |
| rare-word-x5 | 16 | 16 | match |
| repeat-tok | 7 | 7 | match |
| base64 | 21 | 21 | match |
| hex | 3 | 3 | match |
| url | 5 | 5 | match |
| uuid | 22 | 21 | 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 are identical| Field | Inception: Mercury 2.5 | Inception: Mercury 2.5 Preview | Verdict |
|---|---|---|---|
| Context length | 260000 | 260000 | ✓ match |
| Max output tokens | 65536 | 65536 | ✓ match |
| Supported parameters | include_reasoning, max_tokens, reasoning, reasoning_effort, response_format, stop, structured_outputs, temperature, tool_choice, tools | include_reasoning, max_tokens, reasoning, reasoning_effort, response_format, stop, structured_outputs, temperature, tool_choice, tools | ✓ match |
| Defaults | {} | {} | ✓ match |
| Reasoning contract | {"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"medium"} | {"mandatory":false,"default_enabled":true,"supported_efforts":["high","medium","low","none"],"default_effort":"medium"} | ✓ match |
| 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 337 measured models.