@bendyline/gilde 0.1.29 → 0.1.31

package/authoring/chat-models/qwen3.5-4b-q4.json

@@ -13,8 +13,8 @@
     "name": "Alibaba",
     "url": "https://ollama.com/library/qwen3.5"
   },
-  "version": "1.1.2",
-  "updatedAt": "2026-07-28T00:00:00Z",
+  "version": "1.1.3",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "parameterSize": "4B",
   "approxSizeBytes": 2834975040,
@@ -39,8 +39,7 @@
   "llamaCpp": {
     "huggingfaceRepo": "unsloth/Qwen3.5-4B-MTP-GGUF",
     "filename": "Qwen3.5-4B-Q4_K_M.gguf",
-    "quantization": "Q4_K_M",
-    "residentBytes": 3401970048
+    "quantization": "Q4_K_M"
   },
   "releasePatch": {
     "tuning": {

package/authoring/chat-models/qwen3.5-9b-q4.json

@@ -13,8 +13,8 @@
     "name": "Alibaba",
     "url": "https://ollama.com/library/qwen3.5"
   },
-  "version": "1.1.2",
-  "updatedAt": "2026-07-28T00:00:00Z",
+  "version": "1.1.3",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "parameterSize": "9B",
   "approxSizeBytes": 5868826976,
@@ -39,8 +39,7 @@
   "llamaCpp": {
     "huggingfaceRepo": "unsloth/Qwen3.5-9B-MTP-GGUF",
     "filename": "Qwen3.5-9B-Q4_K_M.gguf",
-    "quantization": "Q4_K_M",
-    "residentBytes": 7042592371
+    "quantization": "Q4_K_M"
   },
   "releasePatch": {
     "tuning": {

package/authoring/chat-models/qwen3.6-27b-q4.json

@@ -15,8 +15,8 @@
     "name": "Alibaba",
     "url": "https://ollama.com/library/qwen3.6"
   },
-  "version": "1.1.4",
-  "updatedAt": "2026-07-28T00:00:00Z",
+  "version": "1.1.5",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "recoScore": 20,
   "parameterSize": "27B",
@@ -44,8 +44,7 @@
   "llamaCpp": {
     "huggingfaceRepo": "unsloth/Qwen3.6-27B-MTP-GGUF",
     "filename": "Qwen3.6-27B-Q4_K_M.gguf",
-    "quantization": "Q4_K_M",
-    "residentBytes": 20528127744
+    "quantization": "Q4_K_M"
   },
   "releasePatch": {
     "tuning": {

package/authoring/chat-models/qwen3.6-27b-q8.json

@@ -15,8 +15,8 @@
     "name": "Alibaba",
     "url": "https://ollama.com/library/qwen3.6"
   },
-  "version": "1.0.2",
-  "updatedAt": "2026-07-28T00:00:00Z",
+  "version": "1.0.3",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "recoScore": 20,
   "parameterSize": "27B",
@@ -131,8 +131,7 @@
   "llamaCpp": {
     "huggingfaceRepo": "unsloth/Qwen3.6-27B-MTP-GGUF",
     "filename": "Qwen3.6-27B-Q8_0.gguf",
-    "quantization": "Q8_0",
-    "residentBytes": 34856500992
+    "quantization": "Q8_0"
   },
   "releasePatch": {
     "tuning": {

package/authoring/chat-models/qwen3.6-35b-a3b-q4.json

@@ -16,8 +16,8 @@
     "name": "Alibaba",
     "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"
   },
-  "version": "1.0.1",
-  "updatedAt": "2026-07-28T00:00:00Z",
+  "version": "1.0.2",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "recoScore": 20,
   "parameterSize": "35B",
@@ -132,8 +132,7 @@
   "llamaCpp": {
     "huggingfaceRepo": "unsloth/Qwen3.6-35B-A3B-MTP-GGUF",
     "filename": "Qwen3.6-35B-A3B-UD-Q4_K_M.gguf",
-    "quantization": "UD-Q4_K_M",
-    "residentBytes": 27196064909
+    "quantization": "UD-Q4_K_M"
   },
   "releasePatch": {
     "tuning": {

package/authoring/chat-models/qwen3.6-35b-a3b-q8.json

@@ -16,8 +16,8 @@
     "name": "Alibaba",
     "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"
   },
-  "version": "1.0.1",
-  "updatedAt": "2026-07-28T00:00:00Z",
+  "version": "1.0.2",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "recoScore": 20,
   "parameterSize": "35B",
@@ -132,8 +132,7 @@
   "llamaCpp": {
     "huggingfaceRepo": "unsloth/Qwen3.6-35B-A3B-MTP-GGUF",
     "filename": "Qwen3.6-35B-A3B-Q8_0.gguf",
-    "quantization": "Q8_0",
-    "residentBytes": 45361317005
+    "quantization": "Q8_0"
   },
   "releasePatch": {
     "tuning": {

package/authoring/chat-models/qwen3.8-27b-q4.json

@@ -18,14 +18,14 @@
     "name": "Alibaba",
     "url": "https://huggingface.co/Qwen/Qwen3.8-27B"
   },
-  "version": "1.0.0",
-  "updatedAt": "2026-08-14T00:00:00Z",
+  "version": "1.0.1",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "licenseClass": "open",
   "licenseShortName": "Apache 2.0",
   "recoScore": 20,
   "parameterSize": "27B",
-  "approxSizeBytes": 17106773984,
+  "approxSizeBytes": 17106775008,
   "supportsTools": true,
   "contextWindow": 262144,
   "upstream": "https://huggingface.co/Qwen/Qwen3.8-27B",
@@ -64,9 +64,9 @@
     },
     "reasoning": {
       "enableThinking": true,
-      "thinkingBudget": 8192,
+      "thinkingBudget": 4096,
       "templateKwargs": {
-        "reasoning_effort": "xhigh"
+        "reasoning_effort": "medium"
       }
     },
     "promptTags": {
@@ -86,7 +86,16 @@
         },
         "reasoning": {
           "enableThinking": true,
-          "thinkingBudget": 8192,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "medium"
+          }
+        }
+      },
+      "thinking-deep": {
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
           "templateKwargs": {
             "reasoning_effort": "xhigh"
           }
@@ -104,9 +113,9 @@
         },
         "reasoning": {
           "enableThinking": true,
-          "thinkingBudget": 8192,
+          "thinkingBudget": 4096,
           "templateKwargs": {
-            "reasoning_effort": "xhigh"
+            "reasoning_effort": "medium"
           }
         }
       },
@@ -163,9 +172,13 @@
     "huggingfaceRepo": "unsloth/Qwen3.8-27B-GGUF",
     "filename": "Qwen3.8-27B-Q4_K_M.gguf",
     "quantization": "Q4_K_M",
-    "residentBytes": 20528128781,
+    "residentBytes": 20528130010,
     "mmprojFilename": "mmproj-F16.gguf"
   },
+  "mlx": {
+    "huggingfaceRepo": "mlx-community/Qwen3.8-27B-4bit",
+    "quantization": "4bit"
+  },
   "releasePatch": {
     "tuning": {
       "engine": {

package/authoring/chat-models/qwen3.8-27b-q6.json

@@ -0,0 +1,190 @@
+{
+  "id": "qwen3.8-27b-q6",
+  "name": "Qwen 3.8 (27B, Q6)",
+  "description": "Alibaba's Qwen 3.8 at 27B parameters: a dense, multimodal agentic model with native text, image, and video understanding, 262K context, flexible reasoning effort, MTP, and native tool calling. Q6_K offers a higher-fidelity middle ground between the practical Q4 and larger Q8 builds, targeting machines with 32GB+ usable memory.",
+  "tags": [
+    "alibaba",
+    "agentic",
+    "coding",
+    "multimodal",
+    "vision",
+    "video",
+    "tools",
+    "reasoning",
+    "long-context"
+  ],
+  "category": "general",
+  "maintainer": {
+    "name": "Alibaba",
+    "url": "https://huggingface.co/Qwen/Qwen3.8-27B"
+  },
+  "version": "1.0.0",
+  "updatedAt": "2026-08-15T00:00:00Z",
+  "license": "Apache-2.0",
+  "licenseClass": "open",
+  "licenseShortName": "Apache 2.0",
+  "recoScore": 20,
+  "parameterSize": "27B",
+  "approxSizeBytes": 22884408288,
+  "supportsTools": true,
+  "contextWindow": 262144,
+  "upstream": "https://huggingface.co/Qwen/Qwen3.8-27B",
+  "style": {
+    "family": "qwen",
+    "reasoningFormat": "think",
+    "toolCallFormat": "function-call"
+  },
+  "behaviors": [
+    "reasoning.strip-think-tags",
+    "provider.merge-system-messages",
+    "mcp.compact-tool-schemas",
+    "provider.compact-write-transcript",
+    "fabrication.detect-past-tense-no-tools",
+    "turn.ollama-num-predict-bumped",
+    "turn.preamble-folding",
+    "turn.ramble-detection",
+    "tools.mlx-grammar",
+    "prompt.retrieval-first",
+    "prompt.workspace-gestalt",
+    "prompt.derive-by-execution"
+  ],
+  "tuning": {
+    "sampling": {
+      "temperature": 1,
+      "topP": 0.95,
+      "topK": 20,
+      "minP": 0,
+      "presencePenalty": 0,
+      "repetitionPenalty": 1,
+      "maxTokens": 12288
+    },
+    "samplingWhenThinking": {
+      "temperature": 1,
+      "topP": 0.95
+    },
+    "reasoning": {
+      "enableThinking": true,
+      "thinkingBudget": 4096,
+      "templateKwargs": {
+        "reasoning_effort": "medium"
+      }
+    },
+    "promptTags": {
+      "enableThinkingTag": "/think",
+      "disableThinkingTag": "/no_think"
+    },
+    "profiles": {
+      "thinking-general": {
+        "sampling": {
+          "temperature": 1,
+          "topP": 0.95,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 0,
+          "repetitionPenalty": 1,
+          "maxTokens": 12288
+        },
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "medium"
+          }
+        }
+      },
+      "thinking-deep": {
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "xhigh"
+          }
+        }
+      },
+      "thinking-coding": {
+        "sampling": {
+          "temperature": 1,
+          "topP": 0.95,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 0,
+          "repetitionPenalty": 1,
+          "maxTokens": 16384
+        },
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "medium"
+          }
+        }
+      },
+      "thinking-precise": {
+        "sampling": {
+          "temperature": 0.6,
+          "topP": 0.95,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 0,
+          "repetitionPenalty": 1,
+          "seed": 0,
+          "maxTokens": 6144
+        },
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "medium"
+          }
+        }
+      },
+      "instruct": {
+        "sampling": {
+          "temperature": 0.7,
+          "topP": 0.8,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 1.5,
+          "repetitionPenalty": 1,
+          "maxTokens": 4096
+        },
+        "reasoning": {
+          "enableThinking": false
+        }
+      },
+      "creative": {
+        "sampling": {
+          "temperature": 1,
+          "topP": 0.95,
+          "topK": 40,
+          "minP": 0,
+          "presencePenalty": 1.5,
+          "repetitionPenalty": 1,
+          "maxTokens": 8192
+        },
+        "reasoning": {
+          "enableThinking": false
+        }
+      }
+    }
+  },
+  "llamaCpp": {
+    "huggingfaceRepo": "unsloth/Qwen3.8-27B-GGUF",
+    "filename": "Qwen3.8-27B-Q6_K.gguf",
+    "quantization": "Q6_K",
+    "residentBytes": 27461289946,
+    "mmprojFilename": "mmproj-F16.gguf"
+  },
+  "releasePatch": {
+    "tuning": {
+      "engine": {
+        "llamaCpp": {
+          "spec": {
+            "mtp": true,
+            "nMax": 4
+          }
+        }
+      }
+    }
+  }
+}

package/authoring/chat-models/qwen3.8-27b-q8.json

@@ -0,0 +1,184 @@
+{
+  "id": "qwen3.8-27b-q8",
+  "name": "Qwen 3.8 (27B, Q8)",
+  "description": "Alibaba's Qwen 3.8 at 27B parameters: a dense, multimodal agentic model with native text, image, and video understanding, 262K context, flexible reasoning effort, MTP, and native tool calling. Q8_0 and 8-bit MLX preserve more fidelity than the Q4 build at roughly double the memory footprint; intended for machines with 40GB+ usable memory.",
+  "tags": [
+    "alibaba",
+    "agentic",
+    "coding",
+    "multimodal",
+    "vision",
+    "video",
+    "tools",
+    "reasoning",
+    "long-context"
+  ],
+  "category": "general",
+  "maintainer": {
+    "name": "Alibaba",
+    "url": "https://huggingface.co/Qwen/Qwen3.8-27B"
+  },
+  "version": "1.0.1",
+  "updatedAt": "2026-08-15T00:00:00Z",
+  "license": "Apache-2.0",
+  "licenseClass": "open",
+  "licenseShortName": "Apache 2.0",
+  "recoScore": 20,
+  "parameterSize": "27B",
+  "approxSizeBytes": 29047086048,
+  "supportsTools": true,
+  "contextWindow": 262144,
+  "upstream": "https://huggingface.co/Qwen/Qwen3.8-27B",
+  "style": {
+    "family": "qwen",
+    "reasoningFormat": "think",
+    "toolCallFormat": "function-call"
+  },
+  "behaviors": [
+    "reasoning.strip-think-tags",
+    "provider.merge-system-messages",
+    "mcp.compact-tool-schemas",
+    "provider.compact-write-transcript",
+    "fabrication.detect-past-tense-no-tools",
+    "turn.ollama-num-predict-bumped",
+    "turn.preamble-folding",
+    "turn.ramble-detection",
+    "tools.mlx-grammar",
+    "prompt.retrieval-first",
+    "prompt.workspace-gestalt",
+    "prompt.derive-by-execution"
+  ],
+  "tuning": {
+    "sampling": {
+      "temperature": 1,
+      "topP": 0.95,
+      "topK": 20,
+      "minP": 0,
+      "presencePenalty": 0,
+      "repetitionPenalty": 1,
+      "maxTokens": 12288
+    },
+    "samplingWhenThinking": {
+      "temperature": 1,
+      "topP": 0.95
+    },
+    "reasoning": {
+      "enableThinking": true,
+      "thinkingBudget": 4096,
+      "templateKwargs": {
+        "reasoning_effort": "xhigh"
+      }
+    },
+    "promptTags": {
+      "enableThinkingTag": "/think",
+      "disableThinkingTag": "/no_think"
+    },
+    "profiles": {
+      "thinking-general": {
+        "sampling": {
+          "temperature": 1,
+          "topP": 0.95,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 0,
+          "repetitionPenalty": 1,
+          "maxTokens": 12288
+        },
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "xhigh"
+          }
+        }
+      },
+      "thinking-coding": {
+        "sampling": {
+          "temperature": 1,
+          "topP": 0.95,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 0,
+          "repetitionPenalty": 1,
+          "maxTokens": 16384
+        },
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "xhigh"
+          }
+        }
+      },
+      "thinking-precise": {
+        "sampling": {
+          "temperature": 0.6,
+          "topP": 0.95,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 0,
+          "repetitionPenalty": 1,
+          "seed": 0,
+          "maxTokens": 6144
+        },
+        "reasoning": {
+          "enableThinking": true,
+          "thinkingBudget": 4096,
+          "templateKwargs": {
+            "reasoning_effort": "medium"
+          }
+        }
+      },
+      "instruct": {
+        "sampling": {
+          "temperature": 0.7,
+          "topP": 0.8,
+          "topK": 20,
+          "minP": 0,
+          "presencePenalty": 1.5,
+          "repetitionPenalty": 1,
+          "maxTokens": 4096
+        },
+        "reasoning": {
+          "enableThinking": false
+        }
+      },
+      "creative": {
+        "sampling": {
+          "temperature": 1,
+          "topP": 0.95,
+          "topK": 40,
+          "minP": 0,
+          "presencePenalty": 1.5,
+          "repetitionPenalty": 1,
+          "maxTokens": 8192
+        },
+        "reasoning": {
+          "enableThinking": false
+        }
+      }
+    }
+  },
+  "llamaCpp": {
+    "huggingfaceRepo": "unsloth/Qwen3.8-27B-GGUF",
+    "filename": "Qwen3.8-27B-Q8_0.gguf",
+    "quantization": "Q8_0",
+    "mmprojFilename": "mmproj-F16.gguf"
+  },
+  "mlx": {
+    "huggingfaceRepo": "mlx-community/Qwen3.8-27B-8bit",
+    "quantization": "8bit"
+  },
+  "releasePatch": {
+    "tuning": {
+      "engine": {
+        "llamaCpp": {
+          "spec": {
+            "mtp": true,
+            "nMax": 4
+          }
+        }
+      }
+    }
+  }
+}

package/authoring/chat-models/talkie-1930-13b-q4.json

@@ -14,8 +14,8 @@
     "name": "talkie-lm",
     "url": "https://huggingface.co/talkie-lm/talkie-1930-13b-it"
   },
-  "version": "1.0.0",
-  "updatedAt": "2026-07-20T00:00:00Z",
+  "version": "1.0.1",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "licenseClass": "open",
   "licenseShortName": "Apache 2.0",
@@ -68,7 +68,6 @@
   "llamaCpp": {
     "huggingfaceRepo": "zakarth/talkie-1930-13b-it-vulkan-fixed-GGUF",
     "filename": "talkie-1930-13b-it.Q4_K_M.gguf",
-    "quantization": "Q4_K_M",
-    "residentBytes": 10286533363
+    "quantization": "Q4_K_M"
   }
 }

package/authoring/chat-models/ternary-bonsai-27b-q2.json

@@ -19,8 +19,8 @@
     "name": "PrismML",
     "url": "https://prismml.com"
   },
-  "version": "1.0.0",
-  "updatedAt": "2026-07-14T00:00:00Z",
+  "version": "1.0.1",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "licenseClass": "open",
   "licenseShortName": "Apache 2.0",
@@ -138,7 +138,6 @@
   },
   "mlx": {
     "huggingfaceRepo": "prism-ml/Ternary-Bonsai-27B-mlx-2bit",
-    "quantization": "2bit",
-    "residentBytes": 11100000000
+    "quantization": "2bit"
   }
 }

package/data/chat-models/bt/btl4-35b-q4/manifest.json

@@ -3,7 +3,7 @@
   "kind": "chat-model",
   "id": "btl4-35b-q4",
   "name": "BTL-4 (35B-A3B, Q4)",
-  "description": "Bad Theory Labs' agentic fine-tune of Ornith 1.0 35B, at 35B total parameters with ~2B active per token (Mixture-of-Experts: 256 experts, 8 routed). 30 of its 40 layers use linear attention, so the KV cache costs only ~20 KB/token and the full 256K context stays affordable. Q4_K_M quantization from bartowski — the full-quality companion to the 2.3-bit Compact build. The publisher's benchmark claims are not independently verified.",
+  "description": "Bad Theory Labs' agentic build of Ornith 1.0 35B, tuned for tool-heavy coding and long stretches of unattended work. Native reasoning, native tool calling, and a 256K context window that stays light on memory even when it's full. Capable-tier on-device option for machines with 32GB+ memory — see BTL-4 Compact for a smaller build.",
   "tags": [
     "badtheorylabs",
     "agentic",
@@ -18,8 +18,8 @@
     "name": "Bad Theory Labs",
     "url": "https://huggingface.co/badtheorylabs"
   },
-  "version": "1.0.0",
-  "updatedAt": "2026-08-06T00:00:00Z",
+  "version": "1.0.1",
+  "updatedAt": "2026-08-15T00:00:00Z",
   "license": "Apache-2.0",
   "licenseClass": "open",
   "licenseShortName": "Apache 2.0",
@@ -141,7 +141,6 @@
     "filename": "badtheorylabs_BTL-4-Q4_K_M.gguf",
     "sha256": "75ce22e4f603bdf3f94c2b32151bb6d3492d2ab8a34ad8a4bfa2eaef0125c50f",
     "approxSizeBytes": 21391448192,
-    "quantization": "Q4_K_M",
-    "residentBytes": 25669737830
+    "quantization": "Q4_K_M"
   }
 }

package/data/chat-models/bt/btl4-35b-q4/versions/1.0.1/manifest.json

@@ -0,0 +1,15 @@
+{
+  "schemaVersion": 1,
+  "version": "1.0.1",
+  "releasedAt": "2026-08-15T00:00:00Z",
+  "approxSizeBytes": 21391448192,
+  "llamaCpp": {
+    "huggingfaceRepo": "bartowski/badtheorylabs_BTL-4-GGUF",
+    "revision": "799d011b498e3287f514eb463ea04d6624ec945f",
+    "filename": "badtheorylabs_BTL-4-Q4_K_M.gguf",
+    "sha256": "75ce22e4f603bdf3f94c2b32151bb6d3492d2ab8a34ad8a4bfa2eaef0125c50f",
+    "approxSizeBytes": 21391448192,
+    "quantization": "Q4_K_M"
+  },
+  "minGezelVersion": "1.26227"
+}