modelgrep
A

Arcee AI: Trinity Large Thinking

arcee-ai/trinity-large-thinking

72nd smartest of 178ReasoningToolsJSON
Use via OpenRouter ↗
Intelligence
31.9
72nd of 178
Design Elo
1180
Website
Speed
246
13th fastest
Latency
158ms
first token
Input price
$0.220
104th cheapest
Context
262K
262K max out

How it compares

Smarter than60%
of all ranked models
Faster than96%
of all ranked models
Cheaper than65%
of all ranked models

Overview

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...

Benchmarks

independent · via OpenRouter
Artificial Analysis57th percentile
Intelligence Index
31.9
Coding Index
27.2
Agentic Index
42.6
GPQA Diamond
75%
Humanity's Last Exam
15%
SciCode
36%
Tau²-Bench (agentic)
90%
Design Arena · Elo6,474 tournaments
Website
1180
codecategories
1168
3D
1162
Game Dev
1146
Data Viz
1143
UI Component
1102
asciiart
1084
svg
1073

Providers & pricing (3)

ProviderIn $/MOut $/MUptime
Parasailfp4$0.220$0.850
Arcee AI$0.250$0.800
Venicefp8$0.313$1.13

Specifications

Context window262K
Max output262K
Knowledge cutoff
Input modalitiestext
Output modalitiestext
Prompt caching
Cache read price$0.060/M
ModeratedNo

Trinity Large Thinking FAQ

How much does Trinity Large Thinking cost?

Trinity Large Thinking costs $0.220 per million input tokens and $0.850 per million output tokens via OpenRouter, making it 104th cheapest of 298 paid models.

How smart is Trinity Large Thinking?

Trinity Large Thinking scores 31.9 on the Artificial Analysis Intelligence Index, ranking 72nd of 178 benchmarked models, with a GPQA Diamond score of 75%.

How fast is Trinity Large Thinking?

Trinity Large Thinking generates around 246 tokens per second with 158ms time-to-first-token (p50), the 13th fastest tracked model.

What is Trinity Large Thinking's context window?

Trinity Large Thinking supports a 262K-token context window and can output up to 262K tokens. It accepts text input.

Compare head-to-head