NOUMENOpen app
NOUMEN

arcee · Open weights

Trinity Large Thinking

Added Apr 1, 2026Open weights

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Open source Arcee reasoning model with a 262K context window, 80K max output, and native reasoning and tool support for agentic workloads.

  • Reasoning
  • Tools
Input
Text
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.25
Output
$0.90

Reference prices as published by the model provider. API example

Specifications

Context window
262.1K
Max output
80K
Average output (7 days)
327 tokens
Longer than 35% of 468 models

Side by side

Trinity Large Thinking compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Trinity Large ThinkingarceeThis modelGLM 4.7 FlashZ.aiRemove
ProviderarceeZ.ai
Discovery tierOpen weightsOpen weights
AddedApr 1, 2026Jan 19, 2026
Limits
Context window262.14KBest200K
Max output80K128KBest
Average output327 tokens86 tokens
Pricing, USD per 1M tokens
Input$0.25$0.07Best
Output$0.90$0.40Best
Cache read—$0.01
Performance
Throughput—31.6 tok/s
Time to first token—2.9 s
Request success (24h)—100.0%
Benchmark indexes
Intelligence Index10.8Best10.6
Coding Index25.8—
Capabilities
ReasoningYesYes
ToolsYesYes
Structured outputNoYes
Reasoning effort——

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Trinity Large Thinking
  • GLM 4.7 Flash

Intelligence Index

  • Trinity Large Thinking10.8
  • GLM 4.7 Flash10.6

Trinity Large Thinking: Better than 23% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Trinity Large Thinking: 25.8
    • GLM 4.7 Flash: —

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Trinity Large Thinking: 1.3%
    • GLM 4.7 Flash: —

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Trinity Large Thinking: 0.2%
    • GLM 4.7 Flash: —

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Trinity Large Thinking: 255 Elo
    • GLM 4.7 Flash: —

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Trinity Large Thinking: 336 Elo
    • GLM 4.7 Flash: —

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Trinity Large Thinking: 1.2%
    • GLM 4.7 Flash: —

  • AA-LCRv1.1

    Reasoning across long documents

    • Trinity Large Thinking: 38.0%Top
    • GLM 4.7 Flash: 20.3%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Trinity Large Thinking: 15.8%Top
    • GLM 4.7 Flash: 5.0%

  • GPQA Diamond

    Graduate-level science questions

    • Trinity Large Thinking: 75.2%Top
    • GLM 4.7 Flash: 45.2%

  • IFBench

    Following precise output instructions

    • Trinity Large Thinking: 56.3%Top
    • GLM 4.7 Flash: 46.3%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Trinity Large Thinking: 0.5%
    • GLM 4.7 Flash: —

  • SciCode

    Scientific computing in Python

    • Trinity Large Thinking: 40.6%
    • GLM 4.7 Flash: —

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Trinity Large Thinking: 22.7%Top
    • GLM 4.7 Flash: 3.8%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Trinity Large Thinking: 90.1%
    • GLM 4.7 Flash: 91.8%Top

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Trinity Large Thinking: 38.0%Top
    • GLM 4.7 Flash: 20.3%

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "arcee-ai/trinity-large-thinking",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="arcee-ai/trinity-large-thinking",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)