LLMBOTTLENECK.COM
arcee-ai / afmoe

Trinity-Large-Thinking

398.6 billion parameters, routing 4 of 256 experts per token.

Open in the calculator →

Architecturepublished data
Quick answer
Trinity-Large-Thinking needs about 243.9 GB at Q4_K_M — the format most people download — with 8,192 tokens of context. None of the 18 common devices listed below holds it entirely at this setting. No hardware for it? Rent a GPU that runs it, priced live.

Sized at 8,192 tokens of context with the whole model in device memory, across 18 common devices. Speeds are estimates, not benchmarks.

Weight-file size by format

FormatSizeRangeBasis
FP16792.0 GB782.8 GB – 801.3 GBreconstructed size
Q8_0420.9 GB415.9 GB – 425.8 GBreconstructed size
Q6_K343.2 GBexactpublished data
Q5_K_M283.6 GBexactpublished data
Q5_0272.5 GB269.3 GB – 275.6 GBreconstructed size
Q4_K_M241.8 GBexactpublished data
Q4_0226.3 GBexactpublished data
Q3_K_M181.4 GBexactpublished data
Q2_K139.2 GBexactpublished data

6 of these are published files. Open the pinned GGUF repository ↗

A file size is not the memory a run needs: the KV cache and the runtime reserve come on top, and the calculator adds both.

What the labels mean
published data
Read from a published source — a file's byte count, a model's configuration or a manufacturer's specification — or exact arithmetic on such values. Not a measurement on a machine.
reconstructed size
Weight size reconstructed from the pinned architecture, because no published file exists.
size range
Only a lower and an upper bound are claimed for this weight size.

Hardware ladder

Q4_K_M at 8,192 tokens · 0 of 18 devices hold it entirely
DeviceMemoryVerdictDecodeCalculator
NVIDIA GeForce RTX 5070243.9 GB of 12 GB58 layers on system RAM~9.5 tok/s with offloadOpen →
AMD Radeon™ RX 9070 XT243.9 GB of 16 GB57 layers on system RAM~9.7 tok/s with offloadOpen →
NVIDIA GeForce RTX 4080243.9 GB of 16 GB57 layers on system RAM~9.7 tok/s with offloadOpen →
NVIDIA GeForce RTX 5080243.9 GB of 16 GB57 layers on system RAM~9.8 tok/s with offloadOpen →
NVIDIA GeForce RTX 5070 Ti243.9 GB of 16 GB57 layers on system RAM~9.8 tok/s with offloadOpen →
NVIDIA GeForce RTX 5060 Ti 16GB243.9 GB of 16 GB57 layers on system RAM~9.5 tok/s with offloadOpen →
NVIDIA GeForce RTX 4090243.9 GB of 24 GB55 layers on system RAM~10 tok/s with offloadOpen →
AMD Radeon™ RX 7900 XTX243.9 GB of 24 GB55 layers on system RAM~10 tok/s with offloadOpen →
NVIDIA GeForce RTX 3090243.9 GB of 24 GB55 layers on system RAM~10 tok/s with offloadOpen →
NVIDIA GeForce RTX 5090243.9 GB of 32 GB53 layers on system RAM~11 tok/s with offloadOpen →
NVIDIA RTX 6000 Ada Generation243.9 GB of 48 GB49 layers on system RAM~11 tok/s with offloadOpen →
Apple M4 Pro243.9 GB of 64 GBdoes not fitnot calibratedOpen →
Apple M5 Pro243.9 GB of 64 GBdoes not fitnot calibratedOpen →
AMD Ryzen AI Max+ 395 with Radeon 8060S243.9 GB of 128 GBdoes not fitnot calibratedOpen →
Apple M3 Max243.9 GB of 128 GBdoes not fitnot calibratedOpen →
Apple M4 Max243.9 GB of 128 GBdoes not fitnot calibratedOpen →
NVIDIA DGX Spark243.9 GB of 128 GBdoes not fitnot calibratedOpen →
Apple M2 Ultra243.9 GB of 192 GBdoes not fitnot calibratedOpen →

Decode is a calibrated estimate from the published calibration; capacity is the manufacturer's published ceiling, not guaranteed free memory. Fitting in memory is not the same as loading: whether the runtime and version you have supports this architecture and format on that machine has not been tested here. Offloaded rows assume there is enough system RAM for the overflow — the calculator checks that against the RAM you declare. Each “Open” link keeps this model, format and context.

Run it in the cloud

Rent a GPU that runs Trinity-Large-Thinking

Every rentable machine that holds the whole model, from Vast.ai and RunPod, with this site's speed estimate for each and the provider's own price, read live. Cheapest first; sort by cost per token or speed instead, or switch the format.

Prices · 18:30 UTC, 22 Sept
7 configurations in stock · 9 out of stock
MachineSpeedPer hourPer M tokensWhereRent
8× L40S384 GBrecommendedcheapestbest value~621 tok/sFaster than you read$3.73$1.67Vast.aiMarketplace · 99.9% reliableRent on Vast.ai
8× RTX PRO 5000384 GB~966 tok/sFaster than you read$6.81$1.96Vast.aiMarketplace · 98.3% reliableRent on Vast.ai
2× H200 NVL282 GB~862 tok/sFaster than you read$7.20$2.32Vast.aiMarketplace · 98.7% reliableRent on Vast.ai
1× B300270 GB~691 tok/sFaster than you read$7.89$3.17RunPodSecure CloudRent on RunPod
4× H100 SXM320 GBfastest~1204 tok/sFaster than you read$9.07$2.09Vast.aiMarketplace · 99.9% reliableRunPod $13.96/h Rent on Vast.ai
2× H200282 GB~862 tok/sFaster than you read$9.18$2.96RunPodSecure CloudVast.ai $10.00/h Rent on RunPod
4× H100 PCIe320 GB~718 tok/sFaster than you read$9.59$3.70Vast.aiMarketplace · 98.3% reliableRent on Vast.ai
No GPU at all

Use Trinity-Large-Thinking by the token

For one person chatting, that is about 2.1× cheaper than the best-value rented card above ($1.67 per million tokens). A rented GPU pays off when you keep it busy — many requests batched together, long agent runs — or when the data must stay on a machine you control, or the exact file you want is not served anywhere.

First time renting a GPU? How it works, in four steps
  1. Create an account and add credit. Both providers are prepaid: $10 is enough to try any card on this page for hours. A new RunPod account that arrives through a referral link gets a one-time $5 credit when it first adds $10.
  2. Pick the card and the count shown here, and a template: llama.cpp or vLLM for an OpenAI-compatible API, or one with a web chat if you only want to talk to the model. “How to launch it” below gives the exact image and command.
  3. Wait for the download. The weights are fetched on the machine; tens of gigabytes take a few minutes on a data-centre connection. You pay by the second from the moment it starts.
  4. Stop it when you are done. A running machine bills even when idle. A stopped one costs nothing per hour, though its disk is usually still billed until you delete it.
How to launch Q4_K_M on it (llama.cpp)
Container image
ghcr.io/ggml-org/llama.cpp:server-cuda
Start command / arguments
-hf arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M -c 8192 -ngl 999 --host 0.0.0.0 --port 8080
Expose port
8080 — an OpenAI-compatible API at /v1

Starting from a saved llama.cpp template instead? Leave the arguments empty and set these variables — only the first two change between models:

LLAMA_ARG_HF_REPO=arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M
LLAMA_ARG_CTX_SIZE=8192
LLAMA_ARG_N_GPU_LAYERS=999
LLAMA_ARG_HOST=0.0.0.0
LLAMA_ARG_PORT=8080

Already have the runtime? llama-server -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M -c 8192 -ngl 999 --host 0.0.0.0 --port 8080. Weights: arcee-ai/Trinity-Large-Thinking-GGUF. The CUDA image is for NVIDIA cards; an AMD machine needs llama.cpp's ROCm build.

Every machine holds the whole model at Q4_K_M and 8,192 tokens of context, no offload. Speeds are this site’s single-stream decode estimates; prices are what each provider’s own API quoted, on-demand, for the whole machine. Vast hosts below 98% measured reliability are left out.

Referral links Vast.ai, RunPod and Novita pay us a share of what you spend if you sign up through these buttons. It costs you nothing, and it never decides an order or a recommendation: both are computed from the live price and the speed, and options that pay us nothing are listed and recommended on the same terms. How we rank

Buying a card instead, or paying by the token? Compare a year of Trinity-Large-Thinking three ways →

Under the hood

Architecture, read from the publisher’s file

The numbers every figure above is computed from, with the file they came from.

Architecture

✓ Architecture read from the published config.json

Retrieved 2026-09-01 at pinned commit dc6a99b6b742; parameter count from the safetensors index at the same revision.

Architecture
afmoe
Layers
60
Hidden size
3,072
Attention heads
48
KV heads
8
Head dimension
128
Feed-forward width
12,288
Vocabulary
200,192
Context ceiling
262,144
Sliding window
4,096
RoPE theta
10,000
Experts
256
Experts / token
4
Expert width
3,072
Shared experts
1

Where the memory goes

tokenembedding× 60 decoder blocksGrouped-query attention48 query · 8 KV headswindow 4,096Routed experts4 of 256 per tokenall residentoutputprojectiongrows with contextcapacity ≠ traffic

Grouped-query attention shares each key/value head across 6 query heads, so the KV cache is 17% of what multi-head attention would need at the same context.

Open the exact official config.json ↗

Use this answer in your own product

As JSON, from the API
curl -s https://llmbottleneck.com/v1/analyze \
  -H "Authorization: Bearer $LLMB_KEY" -H "content-type: application/json" \
  -d '{"model":"arcee-ai-trinity-large-thinking","quantization":"Q4_K_M","context":8192,"hardware":"nvidia-geforce-rtx-4090"}'

Same engine, same evidence, every field sourced. Free key in one step →

On your page, as a widget
<script src="https://llmbottleneck.com/widget.js"
  data-model="arcee-ai-trinity-large-thinking" data-quantization="Q4_K_M"
  data-hardware="nvidia-geforce-rtx-4090" data-context="8192"></script>

No key needed. Unbranded, with your own buy button, on Pro and Business →

Citing this page

LLM Bottleneck. “Trinity-Large-Thinking VRAM and hardware requirements.” Architecture from arcee-ai/Trinity-Large-Thinking at revision dc6a99b6b742, retrieved 2026-09-01. https://llmbottleneck.com/models/arcee-ai-trinity-large-thinking

Every figure above is either the published value or a reconstruction whose measured error is on the accuracy page.

llmbottleneck
catalogue 2026-10-03models 327devices 135