Add model card for Qwen3.6-35B-A3B-8bit (#1917)

01598960bd3e19dc1c656e063edaaa3b821d5fb0 · 2026-04-17 13:06:23 +0300 · mlpy0

Adds the 8bit variant missing from #1907 — the safetensors index is now
live on HF.

- `mlx-community/Qwen3.6-35B-A3B-8bit` (~35 GB)

Architectural fields match the existing 4bit/5bit/bf16 cards.
`storage_size.in_bytes` is taken from `metadata.total_size` of the
upstream `model.safetensors.index.json`.

Files touched

A resources/inference_model_cards/mlx-community--Qwen3.6-35B-A3B-8bit.toml

Diff

commit 01598960bd3e19dc1c656e063edaaa3b821d5fb0
Author: mlpy0 <96022931+mlpy0@users.noreply.github.com>
Date:   Fri Apr 17 13:06:23 2026 +0300

    Add model card for Qwen3.6-35B-A3B-8bit (#1917)
    
    Adds the 8bit variant missing from #1907 — the safetensors index is now
    live on HF.
    
    - `mlx-community/Qwen3.6-35B-A3B-8bit` (~35 GB)
    
    Architectural fields match the existing 4bit/5bit/bf16 cards.
    `storage_size.in_bytes` is taken from `metadata.total_size` of the
    upstream `model.safetensors.index.json`.
---
 .../mlx-community--Qwen3.6-35B-A3B-8bit.toml              | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/resources/inference_model_cards/mlx-community--Qwen3.6-35B-A3B-8bit.toml b/resources/inference_model_cards/mlx-community--Qwen3.6-35B-A3B-8bit.toml
new file mode 100644
index 00000000..01b7f671
--- /dev/null
+++ b/resources/inference_model_cards/mlx-community--Qwen3.6-35B-A3B-8bit.toml
@@ -0,0 +1,15 @@
+model_id = "mlx-community/Qwen3.6-35B-A3B-8bit"
+n_layers = 40
+hidden_size = 2048
+num_key_value_heads = 2
+supports_tensor = true
+tasks = ["TextGeneration"]
+family = "qwen"
+quantization = "8bit"
+base_model = "Qwen3.6 35B A3B"
+capabilities = ["text", "thinking", "thinking_toggle", "vision"]
+
+context_length = 262144
+
+[storage_size]
+in_bytes = 37721128672

← 63b8e647 Add model cards for Qwen3.6-35B-A3B variants (#1907) · back to Exo · fix: force gc + clear_cache after KV prefix cache eviction ( af9e847e →