What hardware do you need to run Raon-1 nano?

Raon-1 nano at Q4_K_M is 2.1 GB of weights, with a context ceiling of 256k tokens. The small model in KRAFTON AI's Raon family, post-trained to follow instructions, hold a conversation and call tools in Korean and English, with thinking on by default in its chat template. It is dense, and 1.6B of its 3.1B parameters are a per-layer embedding table: every layer looks up a 256-wide vector for the token in hand and adjusts its own output with it, so a token reads about 1.5B parameters rather than all of them, which is the figure KRAFTON calls effective. It was cut down from Raon-1 mini preview, a 30B-A4B mixture of experts, by pruning and distillation. The 2.08 GB priced here is KRAFTON's own Q4_K_M; the same build with the multi-token-prediction module beside it, for speculative decoding, is 2.15 GB. On the cheapest machine that runs it, a Mac mini M6, 16GB, it generates 37 tok/s at 32k of context, worked out from memory bandwidth.

How good is Raon-1 nano, really?

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

Head to head with a model that needs much the same memory at 32k of context: vs LFM2.5 2.6B.

What Raon-1 nano costs either way

Nobody rents Raon-1 nano by the token. The closest hosted match, Granite 4.2 8B, costs $0.06 per million input tokens and $0.15 per million output (OpenRouter, cheapest active endpoint, checked 2026-11-13). 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. What a million tokens costs each way puts the two prices side by side.

Machines that run Raon-1 nano

Every machine here holds it to 256k, which is this model's own ceiling and the longest the calculator offers, so the choice between them is speed and price rather than how much you can keep in the window.

All 56 machines on this site run it at 32k, not just the eight in the table. One table sets every machine against the others: price, the memory its GPU can use, what it holds and how long it takes to pay for itself.

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 longest context is the longest setting the calculator offers that the machine still holds this model at, cache included, and no machine is shown taking it past its own 256k limit. Each one opens the calculator on that machine at that length.

For scale, the calculator starts from 80 tok/s for a hosted API and times a local machine against it. 5 of the 8 above reach it, and the slowest is 37 tok/s.

The cheapest machine that runs it is sold with 16 GB. There is a page on what 16 GB runs, machine by machine: it names the 6 machines here sold at that size, and the 17 current models the roomiest of them holds at 32k, this one among them.

It fits in less memory than that: the smallest size here that runs it at 32k is 12 GB, on the GeForce RTX 3060, 12GB at $329, card only, which hands a model 11 GB and is no longer sold.

Run Raon-1 nano at 256k on the Mac mini M6, 16GB

How much memory does Raon-1 nano need at 256k?

Every figure above is priced at 32k of context. At 256k, which is this model's own ceiling and the longest the calculator offers, Raon-1 nano needs 3.9 GB rather than 2.3 GB. The weights are the same either way; the key-value cache goes from 0.2 GB to 1.8 GB.

The window costs nothing in hardware: all 37 machines here still sold at a published price hold Raon-1 nano at both lengths, and the cheapest of them is the Mac mini M6, 16GB at $899.

On the Mac mini M6, 16GB it runs 37 tok/s estimated at 32k and 16 tok/s estimated at 256k. Every token generated reads the whole cache back, so a longer window is slower as well as bigger.

The specifics

Parameters

3.1B, of which 1.5B are active per token

Quantisation

Q4_K_M

Weights on disk

2.1 GB

KV cache

0.2 GB at 32k context — Hybrid: 18 of the 24 layers are Raon Delta Attention, which holds a fixed-size state however long the context, and only the 6 latent-attention layers cache a token at all. Each of those keeps one 512-wide compressed vector plus 64 key channels its heads share, 1,152 bytes a token, so 6,912 bytes a token across the six, which rounds to the 7 KiB a token KRAFTON states for this build. The whole 256k window is 1.8 GB of cache against 2.08 GB of weights. How weights and cache add up.

Maximum context

256k tokens

Licence

Apache 2.0

Sources

KRAFTON/Raon-1-nano, KRAFTON/Raon-1-nano (config), KRAFTON/Raon-1-nano-gguf, openrouter.ai