Leaderboard / Arcee AI (Trinity)
Trinity Mini
Arcee AI (Trinity) · Trinity · released 2025-12-01 · data checked 2026-11-01 Needs verification
Trinity Mini is a 26.1B-parameter Mixture-of-Experts model (3B active per token) from Arcee AI (Trinity) (United States), released 2025-12-01 under the OpenMDW-1.1 license. It doesn't have enough reported benchmarks for an index score yet. It needs about 17.1 GB of memory at 4-bit with an 8K-token context.
Small LM Index
—
Needs 2 core benchmarks (has 1)
Parameters
26.1B
3B active · MoE
Price per 1M tokens
—
No hosted API on OpenRouter; self-host
Context window
131K
tokens
Benchmarks
Each bar shows this model's score. The tick marks the best medium model on that benchmark. Sources: 4 reported by Arcee AI (Trinity). How we label sources.
Specifications
- Lab
- Arcee AI (Trinity) (US · United States)
- Architecture
- Mixture-of-Experts
- Total parameters
- 26.1B
- Active parameters
- 3B
- Size class
- Medium
- Context window
- 131K tokens
- Input modalities
- text
- Reasoning
- Reasoning (thinking)
- License
- OpenMDW-1.1
- Openness
- Open weights
- Memory to run
- ~17.1 GB at 4-bit, 8K context (more sizes)
- Hugging Face
- 10.6K downloads (30 days) · 206 likes
Compare with
- Trinity Mini vs DiffusionGemma 26B-A4B (25.8B, index 52.6)
- Trinity Mini vs Qwen3.8 27B (27B, index 77.0)
- Trinity Mini vs Qwen3.6 27B (27B, index 74.0)
Sources
Something wrong here? Report it.
Run it locally
No GGUF build from a well-known publisher yet.
vLLM (GPU server)
vllm serve arcee-ai/Trinity-MiniFull-precision weights, for a server GPU.
Memory needed: weights, context cache and runtime buffers.
| Quantization | 4K context | 32K context | 131K context |
|---|---|---|---|
| 16-bit F16 | 53 GB | 54 GB | 55 GB |
| 8-bit Q8_0 | 29 GB | 29 GB | 31 GB |
| 4-bit Q4_K_M | 17 GB | 17 GB | 19 GB |