Sunk Cost sunkcost.ai Data checked 2026-09-03

What hardware do you need to run Ornith 1.5 9B?

Ornith 1.5 9B at Q4_K_M is 5.8 GB of weights, with a context ceiling of 256k tokens. Not yet rated: released after our last ratings pass. DeepReinforce's MIT-licensed coding model, post-trained on Qwen3.5.

Cheapest machine that runs itMac mini M6, 16GB at $899
Fastest of the ones listedMac Studio M5 Ultra, 96GB — 131 tok/s at 32k context
Honest answer on costPays back in 69 years at 500k tokens a day.

How good is it, really?

It has not been placed on the intelligence index yet. See the whole table.

What it costs either way

Nobody rents Ornith 1.5 9B by the token. The closest hosted match, Qwen3.5 9B, costs $0.08 per million input tokens and $0.13 per million output (OpenRouter, cheapest active endpoint, checked 2026-09-15). Buying a machine only beats that if you use it hard enough, for long enough, that the hardware price divides down below the rental bill.

Machines that run it

MachinePriceSpeed at 32kPay-back
Mac mini M6, 16GB $899 17 tok/s estimated Pays back in 69 years Run the numbers
Strix Halo Framework Desktop, 32GB $1,269 28 tok/s estimated Pays back in 101 years Run the numbers
MacBook Air M5 (13-inch), 16GB $1,299 17 tok/s estimated Pays back in 99 years Run the numbers
MacBook Pro M5 (14-inch), 16GB $1,999 17 tok/s estimated Pays back in 153 years Run the numbers
Mac Studio M5 Max, 36GB $2,499 50 tok/s estimated Pays back in 183 years Run the numbers
DGX Spark GB10 Grace Blackwell, 128GB $4,699 30 tok/s estimated Pays back in 377 years Run the numbers

One machine per family, cheapest first. Speeds are measured where a public benchmark exists and estimated from memory bandwidth otherwise; the calculator says which for any configuration.

The specifics

Parameters
9.7B
Quantisation
Q4_K_M
Weights on disk
5.8 GB
KV cache
1.1 GB at 32k context — Hybrid: three linear-attention layers per full-attention layer. Multimodal; the vision projector is a separate 0.92 GB file, not counted here.
Maximum context
256k tokens (256k)
Licence
MIT
Sources
source 1, source 2