Arcee AI: Trinity Large Thinking

arcee-ai/trinity-large-thinking · Arcee AINot yet established

Declared family: Unknown
Entered catalogue: 1 April 2026
Last tested: 21 August 2026
Measured signature: tk_a8b24bf6

Compare with Cohere: North Mini Code (free)

Identity statement

Arcee AI: Trinity Large Thinking 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_a8b24bf6
How its API is set up	A combination of settings no other listing in the catalogue uses	api_483ad40f
How much it can read and write	Reads up to 262,144 tokens at once, writes up to 80,000	its own listing
Whether it thinks before answering	Always reasons, and cannot be turned off	its own listing
Who served our requests	Parasail	we measured it, 21 August 2026

Closest measured models

Agreement across the mutually clean test strings

Cohere: North Mini Code (free)cohere/north-mini-code:free24/50 · 48%Meta: Llama 4 Scoutmeta-llama/llama-4-scout24/50 · 48%Meta: Llama 4 Maverickmeta-llama/llama-4-maverick24/50 · 48%Meta: Muse Glimmer 30Bmeta/muse-glimmer-30b24/50 · 48%SpaceXAI: Grok Build 0.1x-ai/grok-build-0.124/50 · 48%SpaceXAI: Grok 4.3x-ai/grok-4.324/50 · 48%

Click a row to open the full comparison.

The measured fingerprint

What each test string cost this model, in tokens

We sent Arcee AI: Trinity Large Thinking 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-prose13

en-long15

spaces-203

spaces-603

tabs-203

newlines-204

mixed-ws5

Digits

digits-93

digits-124

digits-3010

digits-sep7

float-long9

CJK

zh-common10

zh-long12

zh-rare19

ja-kana9

ja-kanji11

ko14

Other scripts

ru14

ar16

he17

hi18

th20

el18

Emoji

emoji-basic10

emoji-skin13

emoji-zwj-family7

emoji-zwj-x321

emoji-flags12

emoji-prof14

Rare Unicode

math24

boxdraw21

combining10

cjk-ext-b16

surrogates28

zalgo29

rtl-mix8

Code

py-code25

py-indent15

json20

html16

regex38

camel6

snake6

Repetition and encodings

rare-word-x521

repeat-tok22

base6429

hex10

url17

uuid27

Overhead subtracted: 10 prompt tokens.

Declared record

What the catalogue claims about this model

Trinity Large Thinking is a powerful open source reasoning model from the team at Arcee AI. It shows strong performance in PinchBench, agentic workloads, and reasoning tasks. Launch video: https://youtu.be/Gc82AXLa0Rg?si=4RLn6WBz33qT--B7...

Modalities: text->text
Declared tokenizer: Other
Prompt price: $0.25 / M tokens
Completion price: $0.80 / M tokens

Supported parameters

include_reasoningmax_tokensreasoningtemperaturetool_choicetoolstop_ktop_p

defaults: {"temperature":0.3,"top_p":0.8,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null}

Catalogue entry Weights on Hugging Face

History

Every observation, kept as taken

1 April 2026Enters the OpenRouter catalogue with no declared family.
21 August 2026Fingerprinted in the YFarmX catalogue sweep · 50 of 50 strings measured clean.