LLMBOTTLENECK.COM
ornith-ai / qwen3_5_moe_text

Ornith-1.5-35B-A3B

36.0 billion parameters, routing 8 of 256 experts per token.

Open in the calculator →

Architecturepublished data
Quick answer
Ornith-1.5-35B-A3B needs about 22.7 GB at Q4_K_M — the format most people download — with 8,192 tokens of context. It fits entirely on 12 of 18 common devices, starting with the NVIDIA GeForce RTX 4090 (24 GB). What else fits in 24 GB. 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.

Find the NVIDIA GeForce RTX 4090, the smallest card listed here that holds it: Amazon ↗ · eBay (new and used) ↗Store links may pay us a commission. They never decide which card is suggested — the memory arithmetic does.

Weight-file size by format

FormatSizeRangeBasis
FP1671.1 GBexactpublished data
Q8_037.8 GBexactpublished data
Q6_K29.2 GBexactpublished data
Q5_K_M25.3 GBexactpublished data
Q5_025.1 GB17.3 GB – 31.0 GBsize range
Q4_K_M21.7 GBexactpublished data
Q4_020.9 GB14.2 GB – 31.0 GBsize range
Q3_K_M18.0 GB10.8 GB – 31.0 GBsize range
Q2_K14.2 GB8.26 GB – 31.0 GBsize range

5 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 · 12 of 18 devices hold it entirely
DeviceMemoryVerdictDecodeCalculator
NVIDIA GeForce RTX 507022.7 GB of 12 GB20 layers on system RAM~63 tok/s with offloadOpen →
AMD Radeon™ RX 9070 XT22.7 GB of 16 GB13 layers on system RAM~88 tok/s with offloadOpen →
NVIDIA GeForce RTX 408022.7 GB of 16 GB13 layers on system RAM~87 tok/s with offloadOpen →
NVIDIA GeForce RTX 508022.7 GB of 16 GB13 layers on system RAM~93 tok/s with offloadOpen →
NVIDIA GeForce RTX 5070 Ti22.7 GB of 16 GB13 layers on system RAM~91 tok/s with offloadOpen →
NVIDIA GeForce RTX 5060 Ti 16GB22.7 GB of 16 GB13 layers on system RAM~76 tok/s with offloadOpen →
NVIDIA GeForce RTX 409022.7 GB of 24 GBfits~370 tok/sOpen →
AMD Radeon™ RX 7900 XTX22.7 GB of 24 GBfits~403 tok/sOpen →
NVIDIA GeForce RTX 309022.7 GB of 24 GBfits~343 tok/sOpen →
NVIDIA GeForce RTX 509022.7 GB of 32 GBfits~657 tok/sOpen →
NVIDIA RTX 6000 Ada Generation22.7 GB of 48 GBfits~352 tok/sOpen →
Apple M4 Pro22.7 GB of 64 GBfits~77 tok/sOpen →
Apple M5 Pro22.7 GB of 64 GBfits~82 tok/sOpen →
AMD Ryzen AI Max+ 395 with Radeon 8060S22.7 GB of 128 GBfits~108 tok/sOpen →
Apple M3 Max22.7 GB of 128 GBfits~94 tok/sOpen →
Apple M4 Max22.7 GB of 128 GBfits~107 tok/sOpen →
NVIDIA DGX Spark22.7 GB of 128 GBfits~100 tok/sOpen →
Apple M2 Ultra22.7 GB of 192 GBfits~123 tok/sOpen →

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 Ornith-1.5-35B-A3B

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
26 configurations in stock · 7 out of stock
MachineSpeedPer hourPer M tokensWhereRent
2× RTX 306024 GBcheapest~254 tok/sFaster than you read$0.11$0.12Vast.aiMarketplace · 98.0% reliableRent on Vast.ai
1× RTX 309024 GBrecommendedbest value~343 tok/sFaster than you read$0.12$0.099Vast.aiMarketplace · 99.5% reliableRunPod $0.50/h Rent on Vast.ai
1× RTX 3090 Ti24 GB~369 tok/sFaster than you read$0.19$0.14Vast.aiMarketplace · 99.6% reliableRunPod $0.27/h Rent on Vast.ai
2× RTX 407024 GB~356 tok/sFaster than you read$0.19$0.15Vast.aiMarketplace · 99.8% reliableRent on Vast.ai
2× RTX 4060 Ti 16GB32 GB~203 tok/sFaster than you read$0.22$0.31Vast.aiMarketplace · 99.7% reliableRent on Vast.ai
1× L424 GB~110 tok/sFaster than you read$0.26$0.65Vast.aiMarketplace · 98.9% reliableRunPod $0.49/h Rent on Vast.ai
2× RTX 5060 Ti 16GB32 GB~317 tok/sFaster than you read$0.26$0.22Vast.aiMarketplace · 99.5% reliableRent on Vast.ai
1× RTX 6000 Ada48 GB~352 tok/sFaster than you read$0.26$0.21Vast.aiMarketplace · 99.9% reliableRunPod $0.84/h Rent on Vast.ai

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 ornith-ai/Ornith-1.5-35B-A3B-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=ornith-ai/Ornith-1.5-35B-A3B-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 ornith-ai/Ornith-1.5-35B-A3B-GGUF:Q4_K_M -c 8192 -ngl 999 --host 0.0.0.0 --port 8080. Weights: ornith-ai/Ornith-1.5-35B-A3B-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 Ornith-1.5-35B-A3B 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 10fbf86fed7e; parameter count from the safetensors index at the same revision.

Architecture
qwen3_5_moe_text
Layers
40
Hidden size
2,048
Attention heads
16
KV heads
2
Head dimension
256
Vocabulary
248,320
Context ceiling
262,144
RoPE theta
10,000,000
Experts
256
Experts / token
8
Expert width
512
Shared expert width
512

Where the memory goes

tokenembedding× 40 decoder blocksGrouped-query attention16 query · 2 KV headsfull contextRouted experts8 of 256 per tokenall residentoutputprojectiongrows with contextcapacity ≠ traffic

Grouped-query attention shares each key/value head across 8 query heads, so the KV cache is 13% of what multi-head attention would need at the same context.

This is a multimodal checkpoint (vision); the diagram and memory sizing above cover the text decoder. Encoder/audio/image activations are not included in the KV or weight figures.

Open the exact official config.json ↗

Ornith-1.5-35B-A3B, card by card

One page per device: whether Ornith-1.5-35B-A3B fits, at which formats, and how fast.

More from ornith-ai

Ornith-1.0-35BOrnith-1.0-397BOrnith-1.0-9BOrnith-1.5-397BOrnith-1.5-9B

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":"ornith-ai-ornith-1-5-35b-a3b","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="ornith-ai-ornith-1-5-35b-a3b" 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. “Ornith-1.5-35B-A3B VRAM and hardware requirements.” Architecture from ornith-ai/Ornith-1.5-35B-A3B at revision 10fbf86fed7e, retrieved 2026-09-01. https://llmbottleneck.com/models/ornith-ai-ornith-1-5-35b-a3b

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