Typhoon models
6 Typhoon models from typhoon-ai on Hugging Face, from 2.1B to 30.5B parameters, published by typhoon-ai in BF16, with contexts from 8K to 256K tokens. The smallest official model, typhoon-ocr1.5-2b, needs about 4.8 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 Typhoon size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.
| Model | Parameters | Native VRAM | FP8 VRAM | INT4 VRAM | Live GPU fit (native) | Est. $/hr | KV cache at 32K |
|---|---|---|---|---|---|---|---|
| typhoon-ocr1.5-2b | 2.1B | 4.8 GB | 2.4 GB | 1.2 GB | RTX 3070 | $0.088/hr | 3.5 GB |
| typhoon-ocr-3b | 3.8B | 8.4 GB | 4.2 GB | 2.1 GB | RTX 3060 | $0.110/hr | 1.1 GB |
| typhoon2.5-qwen3-4b | 4.0B | 9.0 GB | 4.5 GB | 2.2 GB | RTX 3060 | $0.110/hr | 4.5 GB |
| llama3.1-typhoon2-8b-instruct | 8.0B | 17.9 GB | 9.0 GB | 4.5 GB | RTX A5000 | $0.176/hr | 4.0 GB |
| typhoon2.5-qwen3-30b-a3b | 30.5B | 68.2 GB | 34.1 GB | 17.1 GB | A100 | $1.18/hr | 3.0 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 (6)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| typhoon-ocr1.5-2b | 2.1B | BF16 | 4.8 GB | RTX 3070 | $0.088/hr |
| typhoon-ocr-3b | 3.8B | BF16 | 8.4 GB | RTX 3060 | $0.110/hr |
| typhoon2.5-qwen3-4b | 4.0B | BF16 | 9.0 GB | RTX 3060 | $0.110/hr |
| llama-3-typhoon-v1.5-8b-instruct | 8.0B | BF16 | 17.9 GB | RTX A5000 | $0.176/hr |
| llama3.1-typhoon2-8b-instruct | 8.0B | BF16 | 17.9 GB | RTX A5000 | $0.176/hr |
| typhoon2.5-qwen3-30b-a3b | 30.5B | BF16 | 68.2 GB | A100 | $1.18/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.