RubyGems - llm.rb - Versions diffs - 16.0.0 → 17.0.1 - Mend

llm.rb 16.0.0 → 17.0.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (88) hide show

checksums.yaml +4 -4
data/CHANGELOG.md +535 -2
data/README.md +857 -501
data/data/alibaba.json +2 -2
data/data/anthropic.json +66 -3
data/data/bedrock.json +781 -12
data/data/deepinfra.json +157 -1
data/data/google.json +48 -0
data/data/mistral.json +326 -16
data/data/moonshot.json +2 -1
data/data/openai.json +13 -0
data/data/openrouter.json +1078 -1235
data/data/xai.json +4 -2
data/docs/deepdive/advanced/cancellation.md +35 -7
data/docs/deepdive/advanced/compaction.md +5 -5
data/docs/deepdive/advanced/context.md +45 -24
data/docs/deepdive/advanced/{guard.md → intercept.md} +87 -83
data/docs/deepdive/features/builtin_tools.md +1 -1
data/docs/deepdive/features/concurrency.md +15 -10
data/docs/deepdive/features/database.md +19 -19
data/docs/deepdive/features/embeddings.md +4 -4
data/docs/deepdive/features/files.md +7 -5
data/docs/deepdive/fundamentals/agents.md +13 -8
data/docs/deepdive/fundamentals/providers.md +1 -1
data/docs/deepdive/fundamentals/schema.md +2 -2
data/docs/deepdive/fundamentals/stream.md +8 -8
data/docs/deepdive/fundamentals/tools.md +14 -14
data/docs/deepdive/media/ocr.md +1 -1
data/docs/deepdive/protocols/a2a.md +3 -3
data/docs/deepdive/protocols/mcp.md +116 -7
data/docs/deepdive/reference/cost.md +1 -1
data/docs/deepdive/reference/response.md +1 -1
data/docs/deepdive/reference/tracer.md +90 -22
data/docs/deepdive.md +1 -1
data/lib/llm/agent/interrupt.rb +35 -46
data/lib/llm/agent/registry.rb +106 -0
data/lib/llm/agent.rb +242 -268
data/lib/llm/context.rb +63 -55
data/lib/llm/function/array.rb +2 -1
data/lib/llm/function/async/group.rb +2 -2
data/lib/llm/function/async/task.rb +25 -48
data/lib/llm/function/fiber/group.rb +2 -2
data/lib/llm/function/fiber/task.rb +16 -30
data/lib/llm/function/fork/group.rb +2 -2
data/lib/llm/function/fork/job.rb +8 -4
data/lib/llm/function/fork/task.rb +25 -30
data/lib/llm/function/group.rb +33 -5
data/lib/llm/function/ractor/group.rb +2 -2
data/lib/llm/function/ractor/job.rb +8 -6
data/lib/llm/function/ractor/task.rb +32 -9
data/lib/llm/function/sequential/group.rb +2 -1
data/lib/llm/function/sequential/task.rb +8 -1
data/lib/llm/function/task.rb +52 -4
data/lib/llm/function/thread/group.rb +2 -2
data/lib/llm/function/thread/task.rb +36 -10
data/lib/llm/function/window.rb +16 -0
data/lib/llm/function.rb +14 -43
data/lib/llm/intercept/multi.rb +40 -0
data/lib/llm/intercept/null.rb +18 -0
data/lib/llm/intercept.rb +96 -0
data/lib/llm/message.rb +1 -1
data/lib/llm/provider.rb +16 -1
data/lib/llm/providers/anthropic/stream_parser.rb +1 -1
data/lib/llm/providers/bedrock/stream_parser.rb +1 -1
data/lib/llm/providers/deepseek/files.rb +49 -0
data/lib/llm/providers/deepseek/request_adapter/completion.rb +15 -13
data/lib/llm/providers/deepseek.rb +16 -2
data/lib/llm/providers/google/stream_parser.rb +1 -1
data/lib/llm/providers/openai/responses/stream_parser.rb +1 -1
data/lib/llm/providers/openai/stream_parser.rb +1 -1
data/lib/llm/stream.rb +8 -7
data/lib/llm/tool.rb +1 -1
data/lib/llm/tools/utils.rb +1 -1
data/lib/llm/tracer/logger.rb +42 -1
data/lib/llm/tracer/multi.rb +144 -0
data/lib/llm/tracer/null.rb +20 -0
data/lib/llm/tracer/pretty_logger.rb +33 -2
data/lib/llm/tracer/rescue.rb +110 -0
data/lib/llm/tracer/telemetry.rb +69 -4
data/lib/llm/tracer.rb +125 -48
data/lib/llm/utils.rb +12 -2
data/lib/llm/version.rb +1 -1
data/lib/llm.rb +36 -1
data/llm.gemspec +9 -4
metadata +16 -35
data/lib/llm/function/tracing.rb +0 -21
data/lib/llm/guard/null.rb +0 -19
data/lib/llm/guard.rb +0 -55

data/data/mistral.json CHANGED Viewed

@@ -94,6 +94,7 @@
         "context": 131072,
         "output": 16384
       },
+      "status": "deprecated",
       "cost": {
         "input": 2,
         "output": 6
@@ -125,6 +126,7 @@
         "context": 128000,
         "output": 128000
       },
+      "status": "deprecated",
       "cost": {
         "input": 0.15,
         "output": 0.15
@@ -181,6 +183,7 @@
         "context": 262144,
         "output": 262144
       },
+      "status": "deprecated",
       "cost": {
         "input": 0.4,
         "output": 2
@@ -250,7 +253,7 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 256000,
+        "context": 262144,
         "output": 256000
       },
       "cost": {
@@ -259,6 +262,39 @@
         "cache_read": 0.015
       }
     },
+    "ministral-8b-2512": {
+      "id": "ministral-8b-2512",
+      "name": "Ministral 3 8B",
+      "description": "Compact open vision-language model for edge deployment, instruction following, and tool use",
+      "family": "ministral",
+      "attachment": true,
+      "reasoning": false,
+      "tool_call": true,
+      "structured_output": true,
+      "temperature": true,
+      "release_date": "2025-12-02",
+      "last_updated": "2025-12-02",
+      "modalities": {
+        "input": [
+          "text",
+          "image"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": true,
+      "limit": {
+        "context": 262144,
+        "output": 262144
+      },
+      "cost": {
+        "input": 0.15,
+        "output": 0.15,
+        "cache_read": 0.015
+      },
+      "canonical_model_id": "mistral/ministral-3-8b-instruct-2512"
+    },
     "zai-glm-5-2": {
       "id": "zai-glm-5-2",
       "name": "GLM-5.2",
@@ -291,10 +327,10 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 1000000,
+        "context": 1048576,
         "output": 131072
       },
-      "status": "beta",
+      "status": "deprecated",
       "cost": {
         "input": 1.4,
         "output": 4.4,
@@ -307,7 +343,7 @@
       "name": "Ministral 8B (latest)",
       "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
       "family": "ministral",
-      "attachment": false,
+      "attachment": true,
       "reasoning": false,
       "tool_call": true,
       "temperature": true,
@@ -316,7 +352,8 @@
       "last_updated": "2024-10-04",
       "modalities": {
         "input": [
-          "text"
+          "text",
+          "image"
         ],
         "output": [
           "text"
@@ -324,7 +361,7 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 128000,
+        "context": 262144,
         "output": 128000
       },
       "cost": {
@@ -363,6 +400,49 @@
         "output": 2
       }
     },
+    "mistral-large-4": {
+      "id": "mistral-large-4",
+      "name": "Mistral Large 4",
+      "description": "Mistral's 1T-parameter multimodal MoE for reasoning, coding, and agentic work",
+      "family": "mistral-large",
+      "attachment": true,
+      "reasoning": true,
+      "reasoning_options": [
+        {
+          "type": "effort",
+          "values": [
+            "none",
+            "high"
+          ]
+        }
+      ],
+      "tool_call": true,
+      "structured_output": true,
+      "temperature": true,
+      "release_date": "2026-10-06",
+      "last_updated": "2026-10-06",
+      "modalities": {
+        "input": [
+          "text",
+          "image"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": false,
+      "limit": {
+        "context": 1048576,
+        "output": 262144
+      },
+      "status": "beta",
+      "cost": {
+        "input": 0.68,
+        "output": 2.09,
+        "cache_read": 0.07
+      },
+      "canonical_model_id": "mistral/mistral-large-4"
+    },
     "open-mixtral-8x22b": {
       "id": "open-mixtral-8x22b",
       "name": "Mixtral 8x22B",
@@ -388,6 +468,7 @@
         "context": 64000,
         "output": 64000
       },
+      "status": "deprecated",
       "cost": {
         "input": 2,
         "output": 6
@@ -450,6 +531,7 @@
         "context": 131072,
         "output": 131072
       },
+      "status": "deprecated",
       "cost": {
         "input": 0.4,
         "output": 2
@@ -460,9 +542,17 @@
       "name": "Magistral Medium (latest)",
       "description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
       "family": "magistral-medium",
-      "attachment": false,
+      "attachment": true,
       "reasoning": true,
-      "reasoning_options": [],
+      "reasoning_options": [
+        {
+          "type": "effort",
+          "values": [
+            "none",
+            "high"
+          ]
+        }
+      ],
       "tool_call": true,
       "temperature": true,
       "knowledge": "2025-06",
@@ -470,7 +560,8 @@
       "last_updated": "2025-03-20",
       "modalities": {
         "input": [
-          "text"
+          "text",
+          "image"
         ],
         "output": [
           "text"
@@ -478,14 +569,129 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 128000,
+        "context": 262144,
         "output": 16384
       },
+      "status": "deprecated",
       "cost": {
         "input": 2,
         "output": 5
       }
     },
+    "voxtral-small-2507": {
+      "id": "voxtral-small-2507",
+      "name": "Voxtral Small",
+      "description": "Instruct model with native audio input for speech understanding and tool use",
+      "family": "voxtral",
+      "attachment": true,
+      "reasoning": false,
+      "tool_call": true,
+      "temperature": true,
+      "release_date": "2025-07-15",
+      "last_updated": "2025-07-15",
+      "modalities": {
+        "input": [
+          "text",
+          "audio"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": true,
+      "limit": {
+        "context": 32768,
+        "output": 32000
+      },
+      "cost": {
+        "input": 0.1,
+        "output": 0.4
+      }
+    },
+    "glm-5-2": {
+      "id": "glm-5-2",
+      "name": "GLM-5.2",
+      "description": "Open flagship GLM for long-horizon coding agents and million-token context work",
+      "family": "glm",
+      "attachment": false,
+      "reasoning": true,
+      "reasoning_options": [
+        {
+          "type": "effort",
+          "values": [
+            "none",
+            "high",
+            "max"
+          ]
+        }
+      ],
+      "tool_call": true,
+      "structured_output": true,
+      "temperature": true,
+      "release_date": "2026-06-13",
+      "last_updated": "2026-06-13",
+      "modalities": {
+        "input": [
+          "text"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": true,
+      "limit": {
+        "context": 1048576,
+        "output": 131072
+      },
+      "status": "deprecated",
+      "cost": {
+        "input": 1.4,
+        "output": 4.4,
+        "cache_read": 0.14
+      },
+      "canonical_model_id": "zhipuai/glm-5.2"
+    },
+    "labs-leanstral-1-5-1": {
+      "id": "labs-leanstral-1-5-1",
+      "name": "Leanstral 1.5",
+      "description": "An updated Lean 4 formal proof engineering model optimised for automated theorem proving and autoformalization. 119B total parameters, 6.5B active.",
+      "family": "leanstral",
+      "attachment": true,
+      "reasoning": true,
+      "reasoning_options": [
+        {
+          "type": "effort",
+          "values": [
+            "none",
+            "high"
+          ]
+        }
+      ],
+      "tool_call": true,
+      "structured_output": true,
+      "temperature": true,
+      "release_date": "2026-06-30",
+      "last_updated": "2026-06-30",
+      "modalities": {
+        "input": [
+          "text",
+          "image"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": true,
+      "limit": {
+        "context": 262144,
+        "output": 128000
+      },
+      "status": "deprecated",
+      "cost": {
+        "input": 0,
+        "output": 0
+      }
+    },
     "pixtral-12b": {
       "id": "pixtral-12b",
       "name": "Pixtral 12B",
@@ -512,6 +718,7 @@
         "context": 128000,
         "output": 128000
       },
+      "status": "deprecated",
       "cost": {
         "input": 0.15,
         "output": 0.15
@@ -609,7 +816,7 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 256000,
+        "context": 262144,
         "output": 256000
       },
       "cost": {
@@ -769,6 +976,39 @@
         "output": 2
       }
     },
+    "ministral-14b-2512": {
+      "id": "ministral-14b-2512",
+      "name": "Ministral 3 14B",
+      "description": "Open vision-language model for efficient local deployment, instruction following, and tool use",
+      "family": "ministral",
+      "attachment": true,
+      "reasoning": false,
+      "tool_call": true,
+      "structured_output": true,
+      "temperature": true,
+      "release_date": "2025-12-02",
+      "last_updated": "2025-12-02",
+      "modalities": {
+        "input": [
+          "text",
+          "image"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": true,
+      "limit": {
+        "context": 262144,
+        "output": 262144
+      },
+      "cost": {
+        "input": 0.2,
+        "output": 0.2,
+        "cache_read": 0.02
+      },
+      "canonical_model_id": "mistral/ministral-3-14b-instruct-2512"
+    },
     "devstral-small-2507": {
       "id": "devstral-small-2507",
       "name": "Devstral Small",
@@ -826,6 +1066,7 @@
         "context": 128000,
         "output": 128000
       },
+      "status": "deprecated",
       "cost": {
         "input": 2,
         "output": 6
@@ -853,7 +1094,7 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 32000,
+        "context": 32768,
         "output": 32000
       },
       "cost": {
@@ -886,6 +1127,7 @@
         "context": 32000,
         "output": 32000
       },
+      "status": "deprecated",
       "cost": {
         "input": 0.7,
         "output": 0.7
@@ -917,6 +1159,7 @@
         "context": 128000,
         "output": 16384
       },
+      "status": "deprecated",
       "cost": {
         "input": 0.1,
         "output": 0.3
@@ -954,7 +1197,7 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 1000000,
+        "context": 1048576,
         "output": 131072
       },
       "cost": {
@@ -964,6 +1207,39 @@
       },
       "canonical_model_id": "zhipuai/glm-5.3"
     },
+    "ministral-3b-2512": {
+      "id": "ministral-3b-2512",
+      "name": "Ministral 3 3B",
+      "description": "Compact open vision-language model for edge deployment, instruction following, and tool use",
+      "family": "ministral",
+      "attachment": true,
+      "reasoning": false,
+      "tool_call": true,
+      "structured_output": true,
+      "temperature": true,
+      "release_date": "2025-12-02",
+      "last_updated": "2025-12-02",
+      "modalities": {
+        "input": [
+          "text",
+          "image"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": true,
+      "limit": {
+        "context": 131072,
+        "output": 262144
+      },
+      "cost": {
+        "input": 0.1,
+        "output": 0.1,
+        "cache_read": 0.01
+      },
+      "canonical_model_id": "mistral/ministral-3-3b-instruct-2512"
+    },
     "mistral-medium-2604": {
       "id": "mistral-medium-2604",
       "name": "Mistral Medium 3.5",
@@ -1030,17 +1306,50 @@
         "context": 8000,
         "output": 8000
       },
+      "status": "deprecated",
       "cost": {
         "input": 0.25,
         "output": 0.25
       }
     },
+    "codestral-2508": {
+      "id": "codestral-2508",
+      "name": "Codestral 25.08",
+      "description": "Mistral code model for low-latency completion, fill-in-the-middle, and code generation",
+      "family": "codestral",
+      "attachment": false,
+      "reasoning": false,
+      "tool_call": true,
+      "structured_output": true,
+      "temperature": true,
+      "release_date": "2025-07-30",
+      "last_updated": "2025-07-30",
+      "modalities": {
+        "input": [
+          "text"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": false,
+      "limit": {
+        "context": 256000,
+        "output": 8192
+      },
+      "cost": {
+        "input": 0.3,
+        "output": 0.9,
+        "cache_read": 0.03
+      },
+      "canonical_model_id": "mistral/codestral-2508"
+    },
     "ministral-3b-latest": {
       "id": "ministral-3b-latest",
       "name": "Ministral 3B (latest)",
       "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
       "family": "ministral",
-      "attachment": false,
+      "attachment": true,
       "reasoning": false,
       "tool_call": true,
       "temperature": true,
@@ -1049,7 +1358,8 @@
       "last_updated": "2024-10-04",
       "modalities": {
         "input": [
-          "text"
+          "text",
+          "image"
         ],
         "output": [
           "text"
@@ -1057,7 +1367,7 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 128000,
+        "context": 131072,
         "output": 128000
       },
       "cost": {

data/data/moonshot.json CHANGED Viewed

@@ -51,7 +51,8 @@
       "cost": {
         "input": 3,
         "output": 15,
-        "cache_read": 0.3
+        "cache_read": 0.3,
+        "cache_write": 3
       },
       "canonical_model_id": "moonshotai/kimi-k3"
     },

data/data/openai.json CHANGED Viewed

@@ -1336,6 +1336,19 @@
                 }
               }
             }
+          },
+          "ultrafast": {
+            "cost": {
+              "input": 12,
+              "output": 60,
+              "cache_read": 0.6,
+              "cache_write": 15
+            },
+            "provider": {
+              "body": {
+                "service_tier": "ultrafast"
+              }
+            }
           }
         }
       },