Skip to content
localmodel.run

Text model · TinyLlama

TLTinyLlama 1.1B requirements

TinyLlama family · 1.1B params · released 2024-01-04 · 3.8M Ollama pulls. Minimum to run at Q4_K_M: Apple M1 (8GB).

LicenseApache-2.0· Commercial OK↓ 1.5M/mo♥ 1.6Kon HuggingFace
Q4_K_M
0.669 GB
Q8_0
1.17 GB
Total @ Q4 (4k)
~1.8 GB
Context
2 k

Quantization sizes

GGUF quantson disk
QuantizationSize on disk
Q2_K0.5 GB est
Q3_K_M0.5 GB est
Q4_K_M (default)0.669 GB
Q5_K_M0.8 GB est
Q6_K0.9 GB est
Q8_01.17 GB
FP162.2 GB est

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

Ollama
$ ollama run tinyllama:1.1b
llama.cpp
$ llama-cli -hf TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF:Q4_K_M
LM Studio
$ lms get TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF

Which devices can run TinyLlama 1.1B?

FAQ

How much VRAM or RAM does TinyLlama 1.1B need?

At Q4_K_M, TinyLlama 1.1B needs about 1.8 GB (weights ~0.669 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~2.3 GB.

Can TinyLlama 1.1B run on a laptop?

Yes, TinyLlama 1.1B fits on a 16 GB machine at Q4_K_M and runs on Apple Silicon or a 12 GB+ GPU comfortably.

Can I use TinyLlama 1.1B commercially?

Yes. TinyLlama 1.1B is licensed Apache-2.0, which permits commercial use.

Community classic 1.1B model pretrained on 3T tokens using Llama-2 architecture. 2048 token context. Very small GGUF footprint (669MB Q4_K_M). Good baseline for edge benchmarking.

Sources

Memory figures are estimates. See methodology.