Snapshot 2026-10-16: flux-3-image-klein matrix (+19 recipes, 1031 total)

README.md

@@ -33,13 +33,13 @@ configs:
 
 Which open-weight AI models actually run on which consumer GPU, and what it takes
 to get them running. 2 727 model×GPU verdicts across 101 models
-and 27 cards, plus 1 012 full setup recipes
-(19 MB of markdown) written against specific hardware.
+and 27 cards, plus 1 031 full setup recipes
+(20 MB of markdown) written against specific hardware.
 
 This is the machine-readable form of [smeltcore.com](https://smeltcore.com). Every row carries a
 `url` back to the page it came from.
 
-Generated 2026-10-16T08:46:27+00:00 from the public read API (`https://api.smeltcore.com/api/v1`) — no private data, no
+Generated 2026-10-16T11:21:15+00:00 from the public read API (`https://api.smeltcore.com/api/v1`) — no private data, no
 credentials, reproducible by anyone.
 
 ## Configs
@@ -47,7 +47,7 @@ credentials, reproducible by anyone.
 | config | rows | what it is |
 |---|---:|---|
 | `compatibility` | 2 727 | **the point of the dataset.** One row per model × GPU, with a verdict |
-| `recipes` | 1 012 | full setup recipes, markdown included, tagged by model / GPU / tool |
+| `recipes` | 1 031 | full setup recipes, markdown included, tagged by model / GPU / tool |
 | `models` | 101 | the catalogue: licence, upstream repo, modality |
 | `gpus` | 27 | the cards: VRAM, vendor, series |
 | `benchmark_sources` | 166 | third-party measurements, normalised and cited |
@@ -65,10 +65,10 @@ The whole dataset turns on this column, so it is worth reading before using it.
 
 | verdict | rows | meaning |
 |---|---:|---|
-| `verified` | 1 011 | somebody ran it on this exact card and wrote down how — there is a recipe behind the row |
-| `fits` | 677 | inferred: the model's memory floor is under the card's VRAM, and the vendor is supported. Not measured |
-| `unknown` | 467 | no floor established for this model, so no honest call can be made |
-| `too_big` | 572 | the memory floor exceeds this card. This is the one verdict asserted from anywhere, not only from same-vendor evidence |
+| `verified` | 1 030 | somebody ran it on this exact card and wrote down how — there is a recipe behind the row |
+| `fits` | 659 | inferred: the model's memory floor is under the card's VRAM, and the vendor is supported. Not measured |
+| `unknown` | 465 | no floor established for this model, so no honest call can be made |
+| `too_big` | 573 | the memory floor exceeds this card. This is the one verdict asserted from anywhere, not only from same-vendor evidence |
 
 The asymmetry between `fits` and `too_big` is deliberate: a model is only called
 runnable on evidence from the same vendor's hardware, but it is called *not*
@@ -125,7 +125,7 @@ extend to the measurements themselves, which belong to the sites named in
   title  = {Self-Hosted AI — GPU Compatibility, Recipes and Catalogue},
   author = {smeltcore},
   url    = {https://smeltcore.com},
-  note   = {Generated 2026-10-16T08:46:27+00:00},
+  note   = {Generated 2026-10-16T11:21:15+00:00},
   year   = {2026}
 }
 ```

data/compatibility.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:587adc03104f25e8355fa94adb0f1a8ebcb75490a2f5793fc812ee18ad8e8cf5
-size 47356
+oid sha256:12c7edc237180442ca78a4c1467baa6587f649e4e53d029c891788ac6a227b16
+size 48308

data/gpus.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:136823b2d7f95e425f5fc5e4635127af35ce48cbf8521e4f0d1ecf4676d23302
-size 3360
+oid sha256:65210e8416cebec8126af36409772d13e1d8cca2138107756c201c8d3cc745f6
+size 3362

data/recipes.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:efdd8b058b0565b88f65a0000fe7ec21d26169908b052df8e58dd505135330e4
-size 3184393
+oid sha256:9964a2307e035376299b64ad70a4a86af80cc43b8925a21ab64c036867f595fd
+size 3206215