Org card: 27B is our first DENSE model (not first single-card), link the showcase Space

README.md

@@ -13,7 +13,8 @@ pinned: false
 
 **A self-funded solo AI research lab. Local-first. Models forged and run daily on hardware we own.**
 
-[![Build log](https://img.shields.io/badge/robinsonlabs.ai-build_log-F59E0B?style=flat-square)](https://robinsonlabs.ai)
+[![Catalog](https://img.shields.io/badge/browse-the_full_catalog-F59E0B?style=flat-square)](https://huggingface.co/spaces/RobinsonLabs/showcase)
+[![Build log](https://img.shields.io/badge/robinsonlabs.ai-build_log-5A5F68?style=flat-square)](https://robinsonlabs.ai)
 [![War stories](https://img.shields.io/badge/writing-war_stories-F59E0B?style=flat-square)](https://robinsonlabs.ai/writing/)
 [![Requests](https://img.shields.io/badge/requests-ask_for_a_model-F59E0B?style=flat-square)](https://huggingface.co/RobinsonLabs/requests/discussions)
 
@@ -41,11 +42,13 @@ The models here are not drive-by uploads. They are built, quantized, and run dai
 
 ## The models
 
-Newest first. Every model ships as a pair: the bf16 safetensors base you can re-quantize or fine-tune from, and the GGUF ladder you can actually run.
+Newest first. Every model ships as a pair: the bf16 safetensors base you can re-quantize or fine-tune from, and the GGUF ladder you can actually run. If you are working on a single card, the 35B-A3B goes smallest: 11.2 GB at IQ2_XS, with 3B active per token.
+
+Every published rung, plotted against size and bits per weight, lives on the [catalog page](https://huggingface.co/spaces/RobinsonLabs/showcase). It is generated from the same build database that produced the files, so it cannot drift from what we actually shipped.
 
 | Family | bf16 base | GGUF ladder | Architecture |
 |---|---|---|---|
-| Qwen3.8-27B abliterated, **vision** | [base](https://huggingface.co/RobinsonLabs/Qwen3.8-27B-abliterated) | [quants](https://huggingface.co/RobinsonLabs/Qwen3.8-27B-abliterated-GGUF) | [view graph](https://hfviewer.com/RobinsonLabs/Qwen3.8-27B-abliterated) |
+| Qwen3.8-27B abliterated, **dense + vision** | [base](https://huggingface.co/RobinsonLabs/Qwen3.8-27B-abliterated) | [quants](https://huggingface.co/RobinsonLabs/Qwen3.8-27B-abliterated-GGUF) | [view graph](https://hfviewer.com/RobinsonLabs/Qwen3.8-27B-abliterated) |
 | Qwen3.5-REAP-262B-A17B abliterated | [base](https://huggingface.co/RobinsonLabs/Qwen3.5-REAP-262B-A17B-abliterated) | [quants](https://huggingface.co/RobinsonLabs/Qwen3.5-REAP-262B-A17B-abliterated-GGUF) | [view graph](https://hfviewer.com/RobinsonLabs/Qwen3.5-REAP-262B-A17B-abliterated) |
 | Qwen3.5-REAP-212B-A17B abliterated | [base](https://huggingface.co/RobinsonLabs/Qwen3.5-REAP-212B-A17B-abliterated) | [quants](https://huggingface.co/RobinsonLabs/Qwen3.5-REAP-212B-A17B-abliterated-GGUF) | [view graph](https://hfviewer.com/RobinsonLabs/Qwen3.5-REAP-212B-A17B-abliterated) |
 | Qwen3.6-35B-A3B abliterated | [base](https://huggingface.co/RobinsonLabs/Qwen3.6-35B-A3B-abliterated) | [quants](https://huggingface.co/RobinsonLabs/Qwen3.6-35B-A3B-abliterated-GGUF) | [view graph](https://hfviewer.com/RobinsonLabs/Qwen3.6-35B-A3B-abliterated) |
@@ -55,7 +58,9 @@ Newest first. Every model ships as a pair: the bf16 safetensors base you can re-
 
 ### New: Qwen3.8-27B abliterated
 
-Our first vision model, and the first one here that a single consumer card can hold. Seven rungs from Q8_0 at 29.05 GB down to IQ3_XS at 12.26 GB, so it fits anything from a 12 GB card upward, with Q4_K_M at 16.84 GB as the volume rung. The f16 `mmproj` ships beside them: download it alongside whichever rung you pick, pass it with `--mmproj`, and you get the image path back. Without it you have a capable text model and no vision.
+Our first pass at a dense model. Everything else in this catalog is a mixture of experts that runs a fraction of its weights on any given token. This one runs all 27 billion every time, which makes it a different tradeoff rather than a smaller box: considerably more compute per token, and a different feel for it. It is also our first with vision.
+
+Seven rungs, from Q8_0 at 29.05 GB down to IQ3_XS at 12.26 GB, with Q4_K_M at 16.84 GB as the volume rung. The f16 `mmproj` ships beside them: download it alongside whichever rung you pick, pass it with `--mmproj`, and you get the image path back. Without it you have a capable text model and no vision.
 
 Two things we did not cut corners on. The vision tower survives abliteration intact. And the MTP block is pinned at or above q5_K on every single rung, never an I-quant, because no importance matrix covers the MTP head. An imatrix is collected during a forward pass that never runs the draft head, so there are no statistics for it, and a low-bit mixture applied blindly there produces garbage. If you re-quantize this model yourself, that is the caveat worth reading first.