Arcee Trinity Mini 26B
Trinity · Apache 2.0
Mixture of Experts: 26B total parameters, 3B active per token.
Arcee AI's mid-range Mixture-of-Experts reasoning model. 26B total parameters with only 3B active per token (128 experts, 8 selected + 1 shared). Trained on 10 trillion tokens via Datology partnership. Handles math, code, and agentic tasks well; fits at Q4_K_M on a 24 GB GPU. The most practical Trinity variant for consumer hardware. US-built, Apache 2.0 licensed.
- Parameters
- 26B
- Architecture
- MoE (3B active)
- Context
- 131,072 tokens
- Released
- 2026-01-27
- Engines
- llama.cpp, ollama, LM Studio, vLLM
- Builder Tools
- Claude Code, Codex CLI, Continue, Cursor, LM Studio, Open WebUI, Windsurf
Parameters
26B
VRAM
20 GB
Context
128K
Formats
4
GPUs
38
Arcee Trinity Mini 26B (26B) requires 20 GB VRAM at recommended quality (Q5_K_M). At efficient quality (Q4_K_M), it fits in 17 GB VRAM, making it compatible with the NVIDIA RTX 4090 Laptop (150-175W). On NVIDIA Grace Blackwell Ultra GB300, expect approximately 332 tok/s at Q8_0. For the best experience, AMD AI Powerhouse ($1,818) is recommended.
Source: OwnRig methodology
20 GB
Q5_K_M
18.6 GB
128K tokens
Chat
VRAM Requirements
| Quality | Quantization | VRAM | File Size |
|---|---|---|---|
| full | Q8_0 | 29 GB | 27.8 GB |
| recommended | Q5_K_M | 20 GB | 18.6 GB |
| efficient | Q4_K_M | 17 GB | 15.9 GB |
| compressed | Q3_K_M | 13 GB | 12.1 GB |
Context Length Impact
KV cache VRAM at Q5_K_M quality. Longer context = more memory.
| Context | KV Cache | Total VRAM |
|---|---|---|
| 2K | 205 MB | 20.2 GB |
| 4K | 307 MB | 20.3 GB |
| 8K | 717 MB | 20.7 GB |
| 16K | 1.4 GB | 21.4 GB |
| 32K | 2.7 GB | 22.7 GB |
| 64K | 5.4 GB | 25.4 GBexceeds 24 GB |
| 128K | 10.9 GB | 30.9 GBexceeds 24 GB |
Compatible GPUs
38 devicesShowing 38 of 38 entries
Builder Context
Arcee Trinity Mini 26B is commonly used with Claude Code, Codex CLI, Continue, Cursor, LM Studio, Open WebUI, Windsurf. For an AI coding workflow, pair it with an embedding model like nomic-embed-text for local RAG.
Frequently Asked Questions
- How much VRAM does Arcee Trinity Mini 26B need?
- Arcee Trinity Mini 26B requires 20 GB VRAM at recommended quality (Q5_K_M). At lower quality settings, it can fit in as little as 13 GB.
- What is the best GPU for Arcee Trinity Mini 26B?
- The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Arcee Trinity Mini 26B, achieving 332 tok/s at Q8_0 with an excellent rating.
- Can I run Arcee Trinity Mini 26B on an RTX 4060 Ti?
- Yes. On the NVIDIA GeForce RTX 4060 Ti 16GB, Arcee Trinity Mini 26B runs at 27 tok/s (Q3_K_M, good).
- What quantization should I use for Arcee Trinity Mini 26B?
- For the best quality, use Q5_K_M (20 GB VRAM). If your GPU has limited VRAM, Q3_K_M (13 GB) is the most efficient option with acceptable quality.
- Is Arcee Trinity Mini 26B good for coding?
- Yes. Arcee Trinity Mini 26B is used with Claude Code, Codex CLI, Continue, Cursor, LM Studio, Open WebUI, Windsurf for local AI coding. For the best coding experience, pair it with an embedding model for local RAG.
Data confidence: estimated. Source
VRAM requirements are calculated from model parameters and may vary by inference engine, context length, and batch size. Performance estimates are based on community benchmarks and should be verified for your specific configuration.Trinity is a trademark of its respective owner. OwnRig is not affiliated with or endorsed by the model creator.