kakaocorp models

3 kakaocorp models from kakaocorp on Hugging Face, from 1.3B to 30.7B parameters, published by kakaocorp in BF16, with a 32K-token context. The smallest official model, kanana-2-1.3b-base, needs about 2.9 GB of VRAM at its published precision; the cheapest live fit is RTX 3070 at $0.088/hr.

Pick a size

One row per official kakaocorp size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.

ModelParametersNative VRAMFP8 VRAMINT4 VRAMLive GPU fit (native)Est. $/hrKV cache at 32K
kanana-2-1.3b-base1.3B2.9 GB1.4 GB0.7 GBRTX 3070$0.088/hr1.1 GB
kanana-2-v-decide-3b-26113.7B8.2 GB4.1 GB2.0 GBRTX A4000$0.167/hrnot published for this architecture
kanana-2-30b-a3b-instruct-260130.7B68.6 GB34.3 GB17.1 GBA100$1.31/hr1.7 GB

VRAM is weights times a flat 1.2 overhead; the KV cache is a separate per-model figure at 16-bit, shown where the architecture is published. See the methodology.

Official models (3)

ModelParametersPublished asVRAM neededLive GPU fitEst. $/hr
kanana-2-1.3b-base1.3BBF162.9 GBRTX 3070$0.088/hr
kanana-2-v-decide-3b-26113.7BBF168.2 GBRTX A4000$0.167/hr
kanana-2-30b-a3b-instruct-260130.7BBF1668.6 GBA100$1.31/hr

VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.

Start ComfyUI on a rented GPU.

© 2026 Aquanode. All rights reserved.

All trademarks, logos and brand names are the property of their respective owners.