How much GPU memory does antares-1b need?

About 6.6 GB for conversations of up to 8,192 tokens: 3.67 GB of weights, the cache for one full-length conversation and the runtime's own memory (an estimate: its settings are behind its licence).

What is the cheapest GPU to run antares-1b?

Right now one RTX 3060 at $0.09 an hour; RTX 5060 Ti at $0.15 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.

Can antares-1b run on an RTX 4090?

Yes: it needs about 6.6 GB and an RTX 4090 has 24 GB.

Models / antares

fdtn-ai/antares-1b

Text generation · served with vLLM · ↓ 3.7K downloads a month · View on Hugging Face

fdtn-ai/antares-1b needs about 6.6 GB of GPU memory for conversations of up to 8,192 tokens (an estimate: its settings are behind its licence). It runs on one RTX 3060 from $0.09/h right now; best value: RTX 5060 Ti at $0.15/h.

The model's publisher asks users to accept its licence (apache-2.0). GridPup cannot download this one yet: choose an ungated version below.

Weights: 3.7 GB
GPU memory needed: about 6.6 GB (estimate)
Conversation length: 8,192 tokens
Memory · disk: 16 GB · 45 GB
Architecture: GraniteMoeHybridForCausalLM
Licence: apache-2.0

Hardware and deploy

The publisher has not let GridPup download this licence-gated model yet. An ungated version of it (above) can be deployed.

Questions

How much GPU memory does antares-1b need?: About 6.6 GB for conversations of up to 8,192 tokens: 3.67 GB of weights, the cache for one full-length conversation and the runtime's own memory (an estimate: its settings are behind its licence).
What is the cheapest GPU to run antares-1b?: Right now one RTX 3060 at $0.09 an hour; RTX 5060 Ti at $0.15 an hour is the best value for its speed. Prices are live offers from GPU clouds, checked every 15 minutes.
Can antares-1b run on an RTX 4090?: Yes: it needs about 6.6 GB and an RTX 4090 has 24 GB.

Loading…