ai/moondream2

Docker

An open-source visual language model that interprets images via text prompts, fast and powerful.

ai/functiongemma-vllm

Docker

FunctionGemma is a 270M open model for fine-tuned, offline function-calling agents on small devices.

ai/granite-embedding-multilingual

Docker

Granite Embedding Multilingual is a 278 million parameter, encoder‑only XLM‑RoBERTa‑style

ai/granite-4.0-h-small

Docker

32B long-context instruct model with RL alignment, IF, tool use, and enterprise optimization.

ai/granite-4.0-h-micro

Docker

3B long-context instruct model with RL alignment, IF, tool calling, and enterprise readiness.

ai/granite-4.0-h-nano

Docker

Granite-4.0-h-nano: lightweight instruct model trained via SFT, RL, and merging on diverse data.

ai/snowflake-arctic-embed-l-v2-vllm

Docker

Snowflake’s Arctic-Embed v2.0 boosts multilingual retrieval and efficiency

ai/embeddinggemma-vllm

Docker

Embedding Gemma is a state-of-the-art text embedding model from Google DeepMind

ai/granite-4.0-micro

Docker

3B long-context instruct model with RL alignment, IF, tool use, and enterprise optimization.

ai/all-minilm-l6-v2-vllm

Docker

all-MiniLM-L6-v2 is a sentence-transformers, maps sentences & paragraphs to a 384 dimensional vector

ai/nomic-embed-text-v2-moe

Docker

Multilingual MoE text embedding model with 768D vectors, 100 languages, 512 token context

aistaging/gemma3

aistaging

Google’s latest Gemma, small yet strong for chat and generation

aistaging/gml-5-safetensors

aistaging

744B MoE language model with 40B active params for reasoning, coding, and agentic tasks (FP8)

aistaging/smollm2-vllm

aistaging

SmolVLM: lightweight multimodal model for video, image, and text analysis, optimized for devices

aistaging/stable-diffusion

aistaging

Image generation model, uses a base latent diffusion model plus a refiner.

aistaging/qwen3-coder

aistaging

30.5B MoE coding model with tool calling, browser automation, and 256K context support

aistaging/ministral3

aistaging

Ministral 3: compact vision-enabled model with near-24B performance, optimized for local edge use

aistaging/minicpm-v-4_5

aistaging

9B multimodal model with vision, speech, and full-duplex streaming for text, image, video, audio

aistaging/translategemma-vllm

aistaging

TranslateGemma: lightweight Gemma 3–based translation models for 55 languages, deployable anywhere

aistaging/gemma3-qat

aistaging

Google’s latest Gemma, in its QAT (quantization aware trained) variant

aistaging/translategemma

aistaging

TranslateGemma: lightweight Gemma 3–based translation models for 55 languages, deployable anywhere

aistaging/ministral3-vllm

aistaging

Ministral 3: compact vision-enabled model with near-24B performance, optimized for local edge use