# Vision-language model support

What the 14 detected VLM families mean in v0.6.2, and what to verify for checkpoint, processor, memory, and serving compatibility.

## Direct answer

The v0.6.2 model library detects exactly 14 VLM model_type families. This is a family-level signal, not universal checkpoint support: image processors, weight variants, quantization, memory use, and serving paths still matter. Eligibility-gated batching remains dense-text-only, so VLM requests stay serial. Check every checkpoint before downloading or exposing it through an API.

- Canonical: https://macmlx.app/models/vision-language-models/
- Last verified: 2026-07-15

## Page facts

- **Fourteen detected VLM families** — Released; since 0.5.3; last verified 2026-07-15. The model library detects 14 vision-language model_type families. This is an evidence-backed family count, not a guarantee that every checkpoint or processor variant will load.
- **Apple Silicon unified memory** — Released; since 0.1.0; last verified 2026-07-15. MLX arrays use the Mac's shared CPU/GPU memory system. Unified memory reduces explicit transfers between CPU orchestration and integrated-GPU compute, but model weights, activations, and KV cache still consume finite physical memory.
- **Eligibility-gated continuous batching** — Released; since 0.6.0; last verified 2026-07-15. v0.6.0 batches only eligible dense-text requests under real concurrency, with automatic serial fallback. The tagged 4-client benchmark measured 2.5–3.2× aggregate throughput. VLM, speculative decoding, Ollama, Anthropic, and embeddings remain serial.

## Sources

- https://github.com/magicnight/mac-mlx/releases/tag/v0.6.2
- https://github.com/magicnight/mac-mlx/blob/v0.6.2/MacMLXCore/Sources/MacMLXCore/Managers/ModelLibraryManager.swift
- https://ml-explore.github.io/mlx/build/html/usage/unified_memory.html
- https://github.com/magicnight/mac-mlx/blob/v0.6.2/MacMLXCore/Sources/MacMLXCore/Engine/MLXSwiftEngine.swift
- https://github.com/magicnight/mac-mlx/blob/v0.6.2/CHANGELOG.md
- https://github.com/magicnight/mac-mlx/blob/v0.6.2/MacMLXCore/Sources/MacMLXCore/Batching/BatchScheduler.swift
- https://github.com/magicnight/mac-mlx/blob/v0.6.2/MacMLXCore/Sources/MacMLXCore/Engine/MLXSwiftEngine+BatchGenerationServing.swift

## Related pages

- [Choose models by task and memory](https://macmlx.app/models/)
- [How to choose an MLX model](https://macmlx.app/models/choosing-a-model/)
- [macMLX v0.6.2](https://macmlx.app/releases/v0-6-2/)