Tasks and inputs
Trinity Large Preview is listed for Text generation. It accepts text and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
SyncDev model directory · Profile updated Oct 13, 2026
Trinity Large Preview is Arcee AI’s custom-licensed sparse MoE checkpoint for general instruction following and agent experiments.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Trinity Large Preview is Arcee AI’s custom-licensed sparse MoE checkpoint for general instruction following and agent experiments.
The publisher describes 398 billion total parameters, about 13 billion active parameters and an extended 512K context. These are different operational quantities: active expert use does not mean the entire checkpoint fits in the memory footprint of a 13B dense model. The release uses the OpenMDW 1.1 licence.
This preview has its own post-training and should be evaluated directly rather than inheriting the Thinking checkpoint’s behaviour. Keep its chat template and serving configuration consistent across trials. A client that silently rewrites history can change a model’s behaviour even when the final user prompt appears identical.
Use a document-based assistant task that requires a clear answer, a tool lookup and an explicit response when a fact is absent. Assess the completed result, the sequence of tool choices and whether the answer remains grounded in supplied material. Include a shorter task and a longer conversation so the evaluation covers ordinary requests as well as the advertised context.
For deployment, verify the expert-parallel serving configuration, chat template and parser support. Record accelerator memory, concurrency and time to the first useful output. Choose a managed route only after checking its exact model identifier and limits; the checkpoint card alone does not establish a universal hosted token price.
Compare the model on the work you need to complete.
Trinity Large Preview is listed for Text generation. It accepts text and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Documented limits: 262.14K context. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
The weights are downloadable under OpenMDW 1.1. Check runtime support and size the complete model and working context for your expected concurrency.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
Anthropic
Claude’s small model for high-volume extraction, classification and routing, with adaptive thinking and a 1M-token context window.
Mistral AI
Mistral Large 4 is an open-weight multimodal model released in public preview on the publisher’s announced release, for general reasoning, coding and agent applications.
OpenAI
GPT-6.1 Sol is OpenAI's updated Sol model for coding and professional tasks. It accepts text and images and returns text.
inclusionAI
Ling 3.1 Flash is inclusionAI’s hybrid reasoning model for coding and tool agents, available through a hosted AI Gateway route.
Anthropic
Claude Sonnet 5.5 is an Anthropic language model that accepts text and images and returns text.
MiniMax
MiniMax M3.1 Flash Preview is a multimodal reasoning model for code and agent workflows, accessed through eligible MiniMax plans and tools.