llm.rb 12.3.1 → 12.5.0

data/data/zai.json

@@ -48,43 +48,6 @@
         "cache_write": 0
       }
     },
-    "glm-4.5v": {
-      "id": "glm-4.5v",
-      "name": "GLM-4.5V",
-      "description": "GLM vision model for visual reasoning, documents, and multimodal agents",
-      "family": "glm",
-      "attachment": true,
-      "reasoning": true,
-      "reasoning_options": [
-        {
-          "type": "toggle"
-        }
-      ],
-      "tool_call": true,
-      "temperature": true,
-      "knowledge": "2025-04",
-      "release_date": "2025-08-11",
-      "last_updated": "2025-08-11",
-      "modalities": {
-        "input": [
-          "text",
-          "image",
-          "video"
-        ],
-        "output": [
-          "text"
-        ]
-      },
-      "open_weights": true,
-      "limit": {
-        "context": 64000,
-        "output": 16384
-      },
-      "cost": {
-        "input": 0.6,
-        "output": 1.8
-      }
-    },
     "glm-4.5": {
       "id": "glm-4.5",
       "name": "GLM-4.5",
@@ -122,11 +85,11 @@
         "cache_write": 0
       }
     },
-    "glm-4.7-flashx": {
-      "id": "glm-4.7-flashx",
-      "name": "GLM-4.7-FlashX",
-      "description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
-      "family": "glm-flash",
+    "glm-5-turbo": {
+      "id": "glm-5-turbo",
+      "name": "GLM-5-Turbo",
+      "description": "Faster GLM-5 lane for coding agents that need lower latency",
+      "family": "glm",
       "attachment": false,
       "reasoning": true,
       "reasoning_options": [
@@ -135,10 +98,13 @@
         }
       ],
       "tool_call": true,
+      "interleaved": {
+        "field": "reasoning_content"
+      },
+      "structured_output": true,
       "temperature": true,
-      "knowledge": "2025-04",
-      "release_date": "2026-01-19",
-      "last_updated": "2026-01-19",
+      "release_date": "2026-03-16",
+      "last_updated": "2026-03-16",
       "modalities": {
         "input": [
           "text"
@@ -147,23 +113,23 @@
           "text"
         ]
       },
-      "open_weights": true,
+      "open_weights": false,
       "limit": {
         "context": 200000,
         "output": 131072
       },
       "cost": {
-        "input": 0.07,
-        "output": 0.4,
-        "cache_read": 0.01,
+        "input": 1.2,
+        "output": 4,
+        "cache_read": 0.24,
         "cache_write": 0
       }
     },
-    "glm-5.1": {
-      "id": "glm-5.1",
-      "name": "GLM-5.1",
-      "description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
-      "family": "glm",
+    "glm-4.7-flashx": {
+      "id": "glm-4.7-flashx",
+      "name": "GLM-4.7-FlashX",
+      "description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
+      "family": "glm-flash",
       "attachment": false,
       "reasoning": true,
       "reasoning_options": [
@@ -172,13 +138,10 @@
         }
       ],
       "tool_call": true,
-      "interleaved": {
-        "field": "reasoning_content"
-      },
-      "structured_output": true,
       "temperature": true,
-      "release_date": "2026-04-07",
-      "last_updated": "2026-04-07",
+      "knowledge": "2025-04",
+      "release_date": "2026-01-19",
+      "last_updated": "2026-01-19",
       "modalities": {
         "input": [
           "text"
@@ -193,17 +156,17 @@
         "output": 131072
       },
       "cost": {
-        "input": 1.4,
-        "output": 4.4,
-        "cache_read": 0.26,
+        "input": 0.07,
+        "output": 0.4,
+        "cache_read": 0.01,
         "cache_write": 0
       }
     },
-    "glm-4.6": {
-      "id": "glm-4.6",
-      "name": "GLM-4.6",
-      "description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
-      "family": "glm",
+    "glm-4.5-air": {
+      "id": "glm-4.5-air",
+      "name": "GLM-4.5-Air",
+      "description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
+      "family": "glm-air",
       "attachment": false,
       "reasoning": true,
       "reasoning_options": [
@@ -214,8 +177,8 @@
       "tool_call": true,
       "temperature": true,
       "knowledge": "2025-04",
-      "release_date": "2025-09-30",
-      "last_updated": "2025-09-30",
+      "release_date": "2025-07-28",
+      "last_updated": "2025-07-28",
       "modalities": {
         "input": [
           "text"
@@ -226,40 +189,33 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 204800,
-        "output": 131072
+        "context": 131072,
+        "output": 98304
       },
       "cost": {
-        "input": 0.6,
-        "output": 2.2,
-        "cache_read": 0.11,
+        "input": 0.2,
+        "output": 1.1,
+        "cache_read": 0.03,
         "cache_write": 0
       }
     },
-    "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",
+    "glm-4.5-flash": {
+      "id": "glm-4.5-flash",
+      "name": "GLM-4.5-Flash",
+      "description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
+      "family": "glm-flash",
       "attachment": false,
       "reasoning": true,
       "reasoning_options": [
         {
-          "type": "effort",
-          "values": [
-            "high",
-            "max"
-          ]
+          "type": "toggle"
         }
       ],
       "tool_call": true,
-      "interleaved": {
-        "field": "reasoning_content"
-      },
-      "structured_output": true,
       "temperature": true,
-      "release_date": "2026-06-13",
-      "last_updated": "2026-06-13",
+      "knowledge": "2025-04",
+      "release_date": "2025-07-28",
+      "last_updated": "2025-07-28",
       "modalities": {
         "input": [
           "text"
@@ -270,13 +226,13 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 1000000,
-        "output": 131072
+        "context": 131072,
+        "output": 98304
       },
       "cost": {
-        "input": 1.4,
-        "output": 4.4,
-        "cache_read": 0.26,
+        "input": 0,
+        "output": 0,
+        "cache_read": 0,
         "cache_write": 0
       }
     },
@@ -317,10 +273,10 @@
         "output": 0.9
       }
     },
-    "glm-5v-turbo": {
-      "id": "glm-5v-turbo",
-      "name": "GLM-5V-Turbo",
-      "description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
+    "glm-4.5v": {
+      "id": "glm-4.5v",
+      "name": "GLM-4.5V",
+      "description": "GLM vision model for visual reasoning, documents, and multimodal agents",
       "family": "glm",
       "attachment": true,
       "reasoning": true,
@@ -330,40 +286,35 @@
         }
       ],
       "tool_call": true,
-      "interleaved": {
-        "field": "reasoning_content"
-      },
       "temperature": true,
-      "release_date": "2026-04-01",
-      "last_updated": "2026-04-01",
+      "knowledge": "2025-04",
+      "release_date": "2025-08-11",
+      "last_updated": "2025-08-11",
       "modalities": {
         "input": [
           "text",
           "image",
-          "video",
-          "pdf"
+          "video"
         ],
         "output": [
           "text"
         ]
       },
-      "open_weights": false,
+      "open_weights": true,
       "limit": {
-        "context": 200000,
-        "output": 131072
+        "context": 64000,
+        "output": 16384
       },
       "cost": {
-        "input": 1.2,
-        "output": 4,
-        "cache_read": 0.24,
-        "cache_write": 0
+        "input": 0.6,
+        "output": 1.8
       }
     },
-    "glm-4.5-air": {
-      "id": "glm-4.5-air",
-      "name": "GLM-4.5-Air",
-      "description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
-      "family": "glm-air",
+    "glm-4.7-flash": {
+      "id": "glm-4.7-flash",
+      "name": "GLM-4.7-Flash",
+      "description": "Budget GLM lane for fast coding help, routing, and everyday automation",
+      "family": "glm-flash",
       "attachment": false,
       "reasoning": true,
       "reasoning_options": [
@@ -374,8 +325,8 @@
       "tool_call": true,
       "temperature": true,
       "knowledge": "2025-04",
-      "release_date": "2025-07-28",
-      "last_updated": "2025-07-28",
+      "release_date": "2026-01-19",
+      "last_updated": "2026-01-19",
       "modalities": {
         "input": [
           "text"
@@ -386,33 +337,40 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 131072,
-        "output": 98304
+        "context": 200000,
+        "output": 131072
       },
       "cost": {
-        "input": 0.2,
-        "output": 1.1,
-        "cache_read": 0.03,
+        "input": 0,
+        "output": 0,
+        "cache_read": 0,
         "cache_write": 0
       }
     },
-    "glm-4.7-flash": {
-      "id": "glm-4.7-flash",
-      "name": "GLM-4.7-Flash",
-      "description": "Budget GLM lane for fast coding help, routing, and everyday automation",
-      "family": "glm-flash",
+    "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": "toggle"
+          "type": "effort",
+          "values": [
+            "high",
+            "max"
+          ]
         }
       ],
       "tool_call": true,
+      "interleaved": {
+        "field": "reasoning_content"
+      },
+      "structured_output": true,
       "temperature": true,
-      "knowledge": "2025-04",
-      "release_date": "2026-01-19",
-      "last_updated": "2026-01-19",
+      "release_date": "2026-06-13",
+      "last_updated": "2026-06-13",
       "modalities": {
         "input": [
           "text"
@@ -423,22 +381,22 @@
       },
       "open_weights": true,
       "limit": {
-        "context": 200000,
+        "context": 1000000,
         "output": 131072
       },
       "cost": {
-        "input": 0,
-        "output": 0,
-        "cache_read": 0,
+        "input": 1.4,
+        "output": 4.4,
+        "cache_read": 0.26,
         "cache_write": 0
       }
     },
-    "glm-4.5-flash": {
-      "id": "glm-4.5-flash",
-      "name": "GLM-4.5-Flash",
-      "description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
-      "family": "glm-flash",
-      "attachment": false,
+    "glm-5v-turbo": {
+      "id": "glm-5v-turbo",
+      "name": "GLM-5V-Turbo",
+      "description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
+      "family": "glm",
+      "attachment": true,
       "reasoning": true,
       "reasoning_options": [
         {
@@ -446,27 +404,32 @@
         }
       ],
       "tool_call": true,
+      "interleaved": {
+        "field": "reasoning_content"
+      },
       "temperature": true,
-      "knowledge": "2025-04",
-      "release_date": "2025-07-28",
-      "last_updated": "2025-07-28",
+      "release_date": "2026-04-01",
+      "last_updated": "2026-04-01",
       "modalities": {
         "input": [
-          "text"
+          "text",
+          "image",
+          "video",
+          "pdf"
         ],
         "output": [
           "text"
         ]
       },
-      "open_weights": true,
+      "open_weights": false,
       "limit": {
-        "context": 131072,
-        "output": 98304
+        "context": 200000,
+        "output": 131072
       },
       "cost": {
-        "input": 0,
-        "output": 0,
-        "cache_read": 0,
+        "input": 1.2,
+        "output": 4,
+        "cache_read": 0.24,
         "cache_write": 0
       }
     },
@@ -509,10 +472,10 @@
         "cache_write": 0
       }
     },
-    "glm-5-turbo": {
-      "id": "glm-5-turbo",
-      "name": "GLM-5-Turbo",
-      "description": "Faster GLM-5 lane for coding agents that need lower latency",
+    "glm-5.1": {
+      "id": "glm-5.1",
+      "name": "GLM-5.1",
+      "description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
       "family": "glm",
       "attachment": false,
       "reasoning": true,
@@ -527,8 +490,8 @@
       },
       "structured_output": true,
       "temperature": true,
-      "release_date": "2026-03-16",
-      "last_updated": "2026-03-16",
+      "release_date": "2026-04-07",
+      "last_updated": "2026-04-07",
       "modalities": {
         "input": [
           "text"
@@ -537,15 +500,52 @@
           "text"
         ]
       },
-      "open_weights": false,
+      "open_weights": true,
       "limit": {
         "context": 200000,
         "output": 131072
       },
       "cost": {
-        "input": 1.2,
-        "output": 4,
-        "cache_read": 0.24,
+        "input": 1.4,
+        "output": 4.4,
+        "cache_read": 0.26,
+        "cache_write": 0
+      }
+    },
+    "glm-4.6": {
+      "id": "glm-4.6",
+      "name": "GLM-4.6",
+      "description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
+      "family": "glm",
+      "attachment": false,
+      "reasoning": true,
+      "reasoning_options": [
+        {
+          "type": "toggle"
+        }
+      ],
+      "tool_call": true,
+      "temperature": true,
+      "knowledge": "2025-04",
+      "release_date": "2025-09-30",
+      "last_updated": "2025-09-30",
+      "modalities": {
+        "input": [
+          "text"
+        ],
+        "output": [
+          "text"
+        ]
+      },
+      "open_weights": true,
+      "limit": {
+        "context": 204800,
+        "output": 131072
+      },
+      "cost": {
+        "input": 0.6,
+        "output": 2.2,
+        "cache_read": 0.11,
         "cache_write": 0
       }
     }

data/lib/llm/agent.rb

@@ -266,6 +266,7 @@ module LLM
     def functions
       @tracer ? @llm.with_tracer(@tracer) { @ctx.functions } : @ctx.functions
     end
+    alias_method :pending_functions, :functions
 
     ##
     # @see LLM::Context#returns
@@ -398,15 +399,18 @@ module LLM
     #  By default this method disables the tracer for
     #  the duration of the repl session, and restores
     #  it afterwards.
-    # @param [Boolean] tracer
-    #  When true, the tracer is kept alive during the
-    #  repl session. Default is false.
+    # @param [String] path
+    #  The path to a file where runtime state is read
+    #  from, and written to
     # @param [Array<LLM::Tool>] tools
     #  Extra tools to attach for the repl session
     # @param [Array<String>] skills
     #  Extra skills to attach for the repl session
+    # @param [Boolean] tracer
+    #  When true, the tracer is kept alive during the
+    #  repl session. Default is false.
     # @return [void]
-    def repl(tracer: false, trace: nil, tools: [], skills: [])
+    def repl(path: nil, tools: [], skills: [], tracer: false, trace: nil)
       if trace != nil
         warn "llm.rb: trace option is deprecated, use tracer instead"
         tracer = trace
@@ -416,9 +420,9 @@ module LLM
         self.tracer = nil
       end
       require_relative "repl" unless defined?(::LLM::Repl)
-      LLM::Repl.new(agent: self, tools:, skills:).start
+      LLM::Repl.new(agent: self, path:, tools:, skills:).start
     ensure
-      if !trace
+      if !tracer
         self.tracer = previous
       end
     end
@@ -460,9 +464,10 @@ module LLM
 
     ##
     # @param (see LLM::Context#deserialize)
-    # @return (see LLM::Context#deserialize)
+    # @return [LLM::Agent]
     def deserialize(**kw)
       @ctx.deserialize(**kw)
+      self
     end
     alias_method :restore, :deserialize
 

data/lib/llm/buffer.rb

@@ -69,6 +69,21 @@ module LLM
       n.nil? ? @messages.last : @messages.last(n)
     end
 
+    ##
+    # Slice a portion of the internal buffer in-place
+    # @return [void]
+    def slice!(...)
+      @messages.slice!(...)
+      nil
+    end
+
+    ##
+    # Pop the last element from the tail of the buffer
+    # @return [void]
+    def pop
+      @messages.pop
+    end
+
     ##
     # @param [[LLM::Message]] item
     #  A message to add to the buffer

data/lib/llm/context/deserializer.rb

@@ -31,9 +31,8 @@ class LLM::Context
     end
     alias_method :restore, :deserialize
 
-    ##
-    # @param [Hash] payload
-    # @return [LLM::Message]
+    private
+
     def deserialize_message(payload)
       tool_calls = deserialize_tool_calls(payload["tools"])
       returns = deserialize_returns(payload["content"]) if returns.nil?
@@ -46,8 +45,6 @@ class LLM::Context
       LLM::Message.new(payload["role"], content, extra)
     end
 
-    private
-
     def deserialize_content(content)
       case content
       when Array

data/lib/llm/context.rb

@@ -257,6 +257,7 @@ module LLM
           end
         end.extend(LLM::Function::Array)
     end
+    alias_method :pending_functions, :functions
 
     ##
     # Returns whether there is pending tool work in this context.
@@ -334,13 +335,11 @@ module LLM
     # This is inspired by Go's context cancellation model.
     # @return [nil]
     def interrupt!
-      pending = functions.to_a
       llm.interrupt!(@owner)
       queue&.interrupt!
-      return if pending.empty?
-      pending.each(&:interrupt!)
-      returns = pending.map { _1.cancel(reason: "function call cancelled") }
-      @messages << LLM::Message.new(@llm.tool_role, returns)
+      functions.each(&:interrupt!)
+      @queue = nil
+      @owner = nil
       nil
     end
     alias_method :cancel!, :interrupt!