llm.rb 13.1.0 → 14.0.0
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.
- checksums.yaml +4 -4
- data/CHANGELOG.md +320 -0
- data/README.md +340 -31
- data/bin/llm.rb +36 -12
- data/data/anthropic.json +206 -263
- data/data/bedrock.json +2138 -1860
- data/data/deepinfra.json +1003 -624
- data/data/deepseek.json +38 -34
- data/data/google.json +1079 -371
- data/data/mistral.json +448 -368
- data/data/moonshot.json +384 -0
- data/data/openai.json +974 -1343
- data/data/xai.json +154 -126
- data/data/zai.json +191 -191
- data/lib/llm/agent.rb +47 -14
- data/lib/llm/context.rb +71 -88
- data/lib/llm/cost.rb +23 -17
- data/lib/llm/error.rb +0 -8
- data/lib/llm/function/async/task.rb +2 -0
- data/lib/llm/function/fiber/task.rb +2 -0
- data/lib/llm/function/fork/task.rb +2 -0
- data/lib/llm/function/ractor/task.rb +2 -0
- data/lib/llm/function/sequential/group.rb +4 -1
- data/lib/llm/function/sequential/task.rb +1 -1
- data/lib/llm/function/task.rb +4 -0
- data/lib/llm/function/thread/task.rb +2 -0
- data/lib/llm/function.rb +32 -4
- data/lib/llm/guard/loop.rb +89 -0
- data/lib/llm/guard/null.rb +19 -0
- data/lib/llm/guard.rb +61 -0
- data/lib/llm/provider.rb +36 -0
- data/lib/llm/providers/anthropic/stream_parser.rb +1 -0
- data/lib/llm/providers/anthropic.rb +1 -8
- data/lib/llm/providers/bedrock/stream_parser.rb +1 -0
- data/lib/llm/providers/bedrock.rb +1 -8
- data/lib/llm/providers/google/stream_parser.rb +1 -0
- data/lib/llm/providers/google.rb +1 -8
- data/lib/llm/providers/moonshot.rb +76 -0
- data/lib/llm/providers/ollama.rb +1 -8
- data/lib/llm/providers/openai/responses/stream_parser.rb +1 -0
- data/lib/llm/providers/openai/responses.rb +6 -8
- data/lib/llm/providers/openai/stream_parser.rb +1 -0
- data/lib/llm/providers/openai.rb +3 -10
- data/lib/llm/repl/bar.rb +4 -3
- data/lib/llm/repl/buffer.rb +42 -15
- data/lib/llm/repl/color.rb +78 -0
- data/lib/llm/repl/input/char.rb +46 -0
- data/lib/llm/repl/input/row.rb +39 -0
- data/lib/llm/repl/input.rb +251 -66
- data/lib/llm/repl/markdown/table.rb +6 -2
- data/lib/llm/repl/markdown.rb +31 -5
- data/lib/llm/repl/status.rb +38 -3
- data/lib/llm/repl/stream.rb +16 -4
- data/lib/llm/repl/walker.rb +3 -2
- data/lib/llm/repl/window.rb +25 -5
- data/lib/llm/repl.rb +29 -13
- data/lib/llm/stream.rb +8 -7
- data/lib/llm/tool.rb +29 -0
- data/lib/llm/transformer/null.rb +21 -0
- data/lib/llm/transformer.rb +55 -0
- data/lib/llm/version.rb +1 -1
- data/lib/llm.rb +12 -2
- data/llm.gemspec +1 -0
- data/resources/deepdive/advanced/cancellation.md +74 -0
- data/resources/deepdive/advanced/compaction.md +83 -0
- data/resources/deepdive/advanced/context.md +267 -0
- data/resources/deepdive/advanced/guard.md +371 -0
- data/resources/deepdive/advanced/tracer.md +180 -0
- data/resources/deepdive/advanced/transformer.md +67 -0
- data/resources/deepdive/advanced/transports.md +45 -0
- data/resources/deepdive/everything_else/audio.md +122 -0
- data/resources/deepdive/everything_else/cost.md +99 -0
- data/resources/deepdive/everything_else/images.md +89 -0
- data/resources/deepdive/everything_else/object.md +108 -0
- data/resources/deepdive/everything_else/ocr.md +48 -0
- data/resources/deepdive/fundamentals/agents.md +202 -0
- data/resources/deepdive/fundamentals/builtin_tools.md +191 -0
- data/resources/deepdive/fundamentals/concurrency.md +104 -0
- data/resources/deepdive/fundamentals/database.md +449 -0
- data/resources/deepdive/fundamentals/embeddings.md +157 -0
- data/resources/deepdive/fundamentals/repl.md +87 -0
- data/resources/deepdive/fundamentals/schema.md +61 -0
- data/resources/deepdive/fundamentals/skills.md +106 -0
- data/resources/deepdive/fundamentals/stream.md +110 -0
- data/resources/deepdive/fundamentals/tools.md +265 -0
- data/resources/deepdive/protocols/a2a.md +106 -0
- data/resources/deepdive/protocols/mcp.md +111 -0
- data/resources/deepdive.md +7 -1
- metadata +36 -3
- data/lib/llm/loop_guard.rb +0 -107
data/data/moonshot.json
ADDED
|
@@ -0,0 +1,384 @@
|
|
|
1
|
+
{
|
|
2
|
+
"id": "moonshotai",
|
|
3
|
+
"env": [
|
|
4
|
+
"MOONSHOT_API_KEY"
|
|
5
|
+
],
|
|
6
|
+
"npm": "@ai-sdk/openai-compatible",
|
|
7
|
+
"api": "https://api.moonshot.ai/v1",
|
|
8
|
+
"name": "Moonshot AI",
|
|
9
|
+
"doc": "https://platform.moonshot.ai/docs/api/chat",
|
|
10
|
+
"models": {
|
|
11
|
+
"kimi-k2.5": {
|
|
12
|
+
"id": "kimi-k2.5",
|
|
13
|
+
"name": "Kimi K2.5",
|
|
14
|
+
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
|
|
15
|
+
"family": "kimi-k2",
|
|
16
|
+
"attachment": false,
|
|
17
|
+
"reasoning": true,
|
|
18
|
+
"reasoning_options": [
|
|
19
|
+
{
|
|
20
|
+
"type": "toggle"
|
|
21
|
+
}
|
|
22
|
+
],
|
|
23
|
+
"tool_call": true,
|
|
24
|
+
"interleaved": {
|
|
25
|
+
"field": "reasoning_content"
|
|
26
|
+
},
|
|
27
|
+
"structured_output": true,
|
|
28
|
+
"temperature": false,
|
|
29
|
+
"knowledge": "2025-01",
|
|
30
|
+
"release_date": "2026-01",
|
|
31
|
+
"last_updated": "2026-01",
|
|
32
|
+
"modalities": {
|
|
33
|
+
"input": [
|
|
34
|
+
"text",
|
|
35
|
+
"image",
|
|
36
|
+
"video"
|
|
37
|
+
],
|
|
38
|
+
"output": [
|
|
39
|
+
"text"
|
|
40
|
+
]
|
|
41
|
+
},
|
|
42
|
+
"open_weights": true,
|
|
43
|
+
"limit": {
|
|
44
|
+
"context": 262144,
|
|
45
|
+
"output": 262144
|
|
46
|
+
},
|
|
47
|
+
"cost": {
|
|
48
|
+
"input": 0.6,
|
|
49
|
+
"output": 3,
|
|
50
|
+
"cache_read": 0.1
|
|
51
|
+
}
|
|
52
|
+
},
|
|
53
|
+
"kimi-k2-thinking-turbo": {
|
|
54
|
+
"id": "kimi-k2-thinking-turbo",
|
|
55
|
+
"name": "Kimi K2 Thinking Turbo",
|
|
56
|
+
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
|
|
57
|
+
"family": "kimi-thinking",
|
|
58
|
+
"attachment": false,
|
|
59
|
+
"reasoning": true,
|
|
60
|
+
"reasoning_options": [],
|
|
61
|
+
"tool_call": true,
|
|
62
|
+
"interleaved": {
|
|
63
|
+
"field": "reasoning_content"
|
|
64
|
+
},
|
|
65
|
+
"temperature": true,
|
|
66
|
+
"knowledge": "2024-08",
|
|
67
|
+
"release_date": "2025-11-06",
|
|
68
|
+
"last_updated": "2025-11-06",
|
|
69
|
+
"modalities": {
|
|
70
|
+
"input": [
|
|
71
|
+
"text"
|
|
72
|
+
],
|
|
73
|
+
"output": [
|
|
74
|
+
"text"
|
|
75
|
+
]
|
|
76
|
+
},
|
|
77
|
+
"open_weights": true,
|
|
78
|
+
"limit": {
|
|
79
|
+
"context": 262144,
|
|
80
|
+
"output": 262144
|
|
81
|
+
},
|
|
82
|
+
"cost": {
|
|
83
|
+
"input": 1.15,
|
|
84
|
+
"output": 8,
|
|
85
|
+
"cache_read": 0.15
|
|
86
|
+
}
|
|
87
|
+
},
|
|
88
|
+
"kimi-k2-0711-preview": {
|
|
89
|
+
"id": "kimi-k2-0711-preview",
|
|
90
|
+
"name": "Kimi K2 0711",
|
|
91
|
+
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
|
|
92
|
+
"family": "kimi-k2",
|
|
93
|
+
"attachment": false,
|
|
94
|
+
"reasoning": false,
|
|
95
|
+
"tool_call": true,
|
|
96
|
+
"temperature": true,
|
|
97
|
+
"knowledge": "2024-10",
|
|
98
|
+
"release_date": "2025-07-14",
|
|
99
|
+
"last_updated": "2025-07-14",
|
|
100
|
+
"modalities": {
|
|
101
|
+
"input": [
|
|
102
|
+
"text"
|
|
103
|
+
],
|
|
104
|
+
"output": [
|
|
105
|
+
"text"
|
|
106
|
+
]
|
|
107
|
+
},
|
|
108
|
+
"open_weights": true,
|
|
109
|
+
"limit": {
|
|
110
|
+
"context": 131072,
|
|
111
|
+
"output": 16384
|
|
112
|
+
},
|
|
113
|
+
"cost": {
|
|
114
|
+
"input": 0.6,
|
|
115
|
+
"output": 2.5,
|
|
116
|
+
"cache_read": 0.15
|
|
117
|
+
}
|
|
118
|
+
},
|
|
119
|
+
"kimi-k2.7-code-highspeed": {
|
|
120
|
+
"id": "kimi-k2.7-code-highspeed",
|
|
121
|
+
"name": "Kimi K2.7 Code HighSpeed",
|
|
122
|
+
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
|
|
123
|
+
"family": "kimi-k2",
|
|
124
|
+
"attachment": true,
|
|
125
|
+
"reasoning": true,
|
|
126
|
+
"reasoning_options": [],
|
|
127
|
+
"tool_call": true,
|
|
128
|
+
"interleaved": {
|
|
129
|
+
"field": "reasoning_content"
|
|
130
|
+
},
|
|
131
|
+
"structured_output": true,
|
|
132
|
+
"temperature": false,
|
|
133
|
+
"knowledge": "2025-01",
|
|
134
|
+
"release_date": "2026-06-12",
|
|
135
|
+
"last_updated": "2026-06-12",
|
|
136
|
+
"modalities": {
|
|
137
|
+
"input": [
|
|
138
|
+
"text",
|
|
139
|
+
"image",
|
|
140
|
+
"video"
|
|
141
|
+
],
|
|
142
|
+
"output": [
|
|
143
|
+
"text"
|
|
144
|
+
]
|
|
145
|
+
},
|
|
146
|
+
"open_weights": true,
|
|
147
|
+
"limit": {
|
|
148
|
+
"context": 262144,
|
|
149
|
+
"output": 262144
|
|
150
|
+
},
|
|
151
|
+
"cost": {
|
|
152
|
+
"input": 1.9,
|
|
153
|
+
"output": 8,
|
|
154
|
+
"cache_read": 0.38
|
|
155
|
+
}
|
|
156
|
+
},
|
|
157
|
+
"kimi-k2.6": {
|
|
158
|
+
"id": "kimi-k2.6",
|
|
159
|
+
"name": "Kimi K2.6",
|
|
160
|
+
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
|
|
161
|
+
"family": "kimi-k2",
|
|
162
|
+
"attachment": true,
|
|
163
|
+
"reasoning": true,
|
|
164
|
+
"reasoning_options": [
|
|
165
|
+
{
|
|
166
|
+
"type": "toggle"
|
|
167
|
+
}
|
|
168
|
+
],
|
|
169
|
+
"tool_call": true,
|
|
170
|
+
"interleaved": {
|
|
171
|
+
"field": "reasoning_content"
|
|
172
|
+
},
|
|
173
|
+
"structured_output": true,
|
|
174
|
+
"temperature": true,
|
|
175
|
+
"knowledge": "2025-01",
|
|
176
|
+
"release_date": "2026-04-21",
|
|
177
|
+
"last_updated": "2026-04-21",
|
|
178
|
+
"modalities": {
|
|
179
|
+
"input": [
|
|
180
|
+
"text",
|
|
181
|
+
"image",
|
|
182
|
+
"video"
|
|
183
|
+
],
|
|
184
|
+
"output": [
|
|
185
|
+
"text"
|
|
186
|
+
]
|
|
187
|
+
},
|
|
188
|
+
"open_weights": true,
|
|
189
|
+
"limit": {
|
|
190
|
+
"context": 262144,
|
|
191
|
+
"output": 262144
|
|
192
|
+
},
|
|
193
|
+
"cost": {
|
|
194
|
+
"input": 0.95,
|
|
195
|
+
"output": 4,
|
|
196
|
+
"cache_read": 0.16
|
|
197
|
+
}
|
|
198
|
+
},
|
|
199
|
+
"kimi-k2-turbo-preview": {
|
|
200
|
+
"id": "kimi-k2-turbo-preview",
|
|
201
|
+
"name": "Kimi K2 Turbo",
|
|
202
|
+
"description": "Fast Kimi model for responsive chat, coding help, and agent loops",
|
|
203
|
+
"family": "kimi-k2",
|
|
204
|
+
"attachment": false,
|
|
205
|
+
"reasoning": false,
|
|
206
|
+
"tool_call": true,
|
|
207
|
+
"temperature": true,
|
|
208
|
+
"knowledge": "2024-10",
|
|
209
|
+
"release_date": "2025-09-05",
|
|
210
|
+
"last_updated": "2025-09-05",
|
|
211
|
+
"modalities": {
|
|
212
|
+
"input": [
|
|
213
|
+
"text"
|
|
214
|
+
],
|
|
215
|
+
"output": [
|
|
216
|
+
"text"
|
|
217
|
+
]
|
|
218
|
+
},
|
|
219
|
+
"open_weights": true,
|
|
220
|
+
"limit": {
|
|
221
|
+
"context": 262144,
|
|
222
|
+
"output": 262144
|
|
223
|
+
},
|
|
224
|
+
"cost": {
|
|
225
|
+
"input": 2.4,
|
|
226
|
+
"output": 10,
|
|
227
|
+
"cache_read": 0.6
|
|
228
|
+
}
|
|
229
|
+
},
|
|
230
|
+
"kimi-k2-0905-preview": {
|
|
231
|
+
"id": "kimi-k2-0905-preview",
|
|
232
|
+
"name": "Kimi K2 0905",
|
|
233
|
+
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
|
|
234
|
+
"family": "kimi-k2",
|
|
235
|
+
"attachment": false,
|
|
236
|
+
"reasoning": false,
|
|
237
|
+
"tool_call": true,
|
|
238
|
+
"temperature": true,
|
|
239
|
+
"knowledge": "2024-10",
|
|
240
|
+
"release_date": "2025-09-05",
|
|
241
|
+
"last_updated": "2025-09-05",
|
|
242
|
+
"modalities": {
|
|
243
|
+
"input": [
|
|
244
|
+
"text"
|
|
245
|
+
],
|
|
246
|
+
"output": [
|
|
247
|
+
"text"
|
|
248
|
+
]
|
|
249
|
+
},
|
|
250
|
+
"open_weights": true,
|
|
251
|
+
"limit": {
|
|
252
|
+
"context": 262144,
|
|
253
|
+
"output": 262144
|
|
254
|
+
},
|
|
255
|
+
"cost": {
|
|
256
|
+
"input": 0.6,
|
|
257
|
+
"output": 2.5,
|
|
258
|
+
"cache_read": 0.15
|
|
259
|
+
}
|
|
260
|
+
},
|
|
261
|
+
"kimi-k2.7-code": {
|
|
262
|
+
"id": "kimi-k2.7-code",
|
|
263
|
+
"name": "Kimi K2.7 Code",
|
|
264
|
+
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
|
|
265
|
+
"family": "kimi-k2",
|
|
266
|
+
"attachment": true,
|
|
267
|
+
"reasoning": true,
|
|
268
|
+
"reasoning_options": [],
|
|
269
|
+
"tool_call": true,
|
|
270
|
+
"interleaved": {
|
|
271
|
+
"field": "reasoning_content"
|
|
272
|
+
},
|
|
273
|
+
"structured_output": true,
|
|
274
|
+
"temperature": false,
|
|
275
|
+
"knowledge": "2025-01",
|
|
276
|
+
"release_date": "2026-06-12",
|
|
277
|
+
"last_updated": "2026-06-12",
|
|
278
|
+
"modalities": {
|
|
279
|
+
"input": [
|
|
280
|
+
"text",
|
|
281
|
+
"image",
|
|
282
|
+
"video"
|
|
283
|
+
],
|
|
284
|
+
"output": [
|
|
285
|
+
"text"
|
|
286
|
+
]
|
|
287
|
+
},
|
|
288
|
+
"open_weights": true,
|
|
289
|
+
"limit": {
|
|
290
|
+
"context": 262144,
|
|
291
|
+
"output": 262144
|
|
292
|
+
},
|
|
293
|
+
"cost": {
|
|
294
|
+
"input": 0.95,
|
|
295
|
+
"output": 4,
|
|
296
|
+
"cache_read": 0.19
|
|
297
|
+
}
|
|
298
|
+
},
|
|
299
|
+
"kimi-k2-thinking": {
|
|
300
|
+
"id": "kimi-k2-thinking",
|
|
301
|
+
"name": "Kimi K2 Thinking",
|
|
302
|
+
"description": "Thinking Kimi model for slower research passes, planning, and hard technical questions",
|
|
303
|
+
"family": "kimi-thinking",
|
|
304
|
+
"attachment": false,
|
|
305
|
+
"reasoning": true,
|
|
306
|
+
"reasoning_options": [],
|
|
307
|
+
"tool_call": true,
|
|
308
|
+
"interleaved": {
|
|
309
|
+
"field": "reasoning_content"
|
|
310
|
+
},
|
|
311
|
+
"temperature": true,
|
|
312
|
+
"knowledge": "2024-08",
|
|
313
|
+
"release_date": "2025-11-06",
|
|
314
|
+
"last_updated": "2025-11-06",
|
|
315
|
+
"modalities": {
|
|
316
|
+
"input": [
|
|
317
|
+
"text"
|
|
318
|
+
],
|
|
319
|
+
"output": [
|
|
320
|
+
"text"
|
|
321
|
+
]
|
|
322
|
+
},
|
|
323
|
+
"open_weights": true,
|
|
324
|
+
"limit": {
|
|
325
|
+
"context": 262144,
|
|
326
|
+
"output": 262144
|
|
327
|
+
},
|
|
328
|
+
"cost": {
|
|
329
|
+
"input": 0.6,
|
|
330
|
+
"output": 2.5,
|
|
331
|
+
"cache_read": 0.15
|
|
332
|
+
}
|
|
333
|
+
},
|
|
334
|
+
"kimi-k3": {
|
|
335
|
+
"id": "kimi-k3",
|
|
336
|
+
"name": "Kimi K3",
|
|
337
|
+
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
|
|
338
|
+
"family": "kimi-k3",
|
|
339
|
+
"attachment": true,
|
|
340
|
+
"reasoning": true,
|
|
341
|
+
"reasoning_options": [
|
|
342
|
+
{
|
|
343
|
+
"type": "toggle"
|
|
344
|
+
},
|
|
345
|
+
{
|
|
346
|
+
"type": "effort",
|
|
347
|
+
"values": [
|
|
348
|
+
"low",
|
|
349
|
+
"high",
|
|
350
|
+
"max"
|
|
351
|
+
]
|
|
352
|
+
}
|
|
353
|
+
],
|
|
354
|
+
"tool_call": true,
|
|
355
|
+
"interleaved": {
|
|
356
|
+
"field": "reasoning_content"
|
|
357
|
+
},
|
|
358
|
+
"structured_output": true,
|
|
359
|
+
"temperature": false,
|
|
360
|
+
"release_date": "2026-07-16",
|
|
361
|
+
"last_updated": "2026-07-16",
|
|
362
|
+
"modalities": {
|
|
363
|
+
"input": [
|
|
364
|
+
"text",
|
|
365
|
+
"image",
|
|
366
|
+
"video"
|
|
367
|
+
],
|
|
368
|
+
"output": [
|
|
369
|
+
"text"
|
|
370
|
+
]
|
|
371
|
+
},
|
|
372
|
+
"open_weights": true,
|
|
373
|
+
"limit": {
|
|
374
|
+
"context": 1048576,
|
|
375
|
+
"output": 131072
|
|
376
|
+
},
|
|
377
|
+
"cost": {
|
|
378
|
+
"input": 3,
|
|
379
|
+
"output": 15,
|
|
380
|
+
"cache_read": 0.3
|
|
381
|
+
}
|
|
382
|
+
}
|
|
383
|
+
}
|
|
384
|
+
}
|