Can I run Trinity-Large-TrueBase?

Trinity-Large-TrueBase needs about 278.9 GB of memory at Q4_K_M, so it does not fit any single consumer GPU we track. The cheapest Mac or mini PC that holds it is the Mac Studio M5 Ultra 512GB (512 GB, $22299). 1 of the 82 devices we track run it fully at Q4. These are estimates, not measured benchmarks.

398.6B parameters, by arcee-ai. License: other. Model card (params from HF API, retrieved 2026-11-04) · How much VRAM for 300B–480B models · Q4 vs Q8 vs FP16 · MoE: 13.0B active per token

Q4_K_M needs

278.9 GB 241.7 GB weights

Q8_0 needs

488.1 GB 423.5 GB weights

FP16 needs

917.9 GB 797.3 GB weights

1 of 82 tracked GPUs and Macs run it fully at Q4.

Memory by context length

Longer chats and documents need more memory for the KV cache. Only 15 of its 60 layers keep a cache that grows with context (45 use a 4096-token sliding window), so it stays lean at long contexts. Sizes assume an FP16 KV cache; a Q8 cache roughly halves the cache part.

From the model's config.json (15 full-attention layers × 8 KV heads × 128 head size, retrieved 2026-11-04). Check the model card for its maximum supported context. Try other sizes in the VRAM calculator or compare models in our context length guide.

Which hardware runs it

We may earn a commission when you buy through these links, at no extra cost to you.