@hermespilot/link 0.7.9-beta.0 → 0.7.9
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.
- package/.env.example +27 -0
- package/README.md +2 -0
- package/dist/{chunk-Y5S7U77X.js → chunk-LUGNDJU3.js} +3128 -528
- package/dist/cli/index.js +1 -1
- package/dist/http/app.d.ts +52 -0
- package/dist/http/app.js +1 -1
- package/model-capabilities/catalog.json +1228 -0
- package/package.json +3 -1
|
@@ -0,0 +1,1228 @@
|
|
|
1
|
+
{
|
|
2
|
+
"schemaVersion": 1,
|
|
3
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
4
|
+
"templateGuide": {
|
|
5
|
+
"purpose": "Template only. Replace the sample rows with verified model capability records before publishing to R2.",
|
|
6
|
+
"matching": "Use canonicalId only. Link performs conservative normalization when matching model IDs.",
|
|
7
|
+
"reasoningSupportPolicy": {
|
|
8
|
+
"supported": "Use when official/provider docs clearly say this model supports reasoning controls.",
|
|
9
|
+
"unsupported": "Use when official/provider docs clearly say this model does not support reasoning controls.",
|
|
10
|
+
"null": "Use when reasoning support has not been verified. Do not write auto in the catalog; auto is only a Profile/App configuration policy."
|
|
11
|
+
},
|
|
12
|
+
"supportedReasoningEfforts": {
|
|
13
|
+
"allowedValues": ["none", "minimal", "low", "medium", "high", "xhigh"],
|
|
14
|
+
"forSupportedModels": "Use the exact verified subset. The sample reasoning row shows the complete possible list.",
|
|
15
|
+
"forUnsupportedModels": [],
|
|
16
|
+
"forUnknownModels": null
|
|
17
|
+
}
|
|
18
|
+
},
|
|
19
|
+
"models": [
|
|
20
|
+
{
|
|
21
|
+
"canonicalId": "deepseek-v4-flash",
|
|
22
|
+
"contextLength": 1000000,
|
|
23
|
+
"maxOutputTokens": 384000,
|
|
24
|
+
"reasoningSupportPolicy": "supported",
|
|
25
|
+
"supportedReasoningEfforts": [
|
|
26
|
+
"low",
|
|
27
|
+
"medium",
|
|
28
|
+
"high",
|
|
29
|
+
"xhigh"
|
|
30
|
+
],
|
|
31
|
+
"supportsVision": null,
|
|
32
|
+
"inputModalities": ["text"],
|
|
33
|
+
"outputModalities": ["text"],
|
|
34
|
+
"source": "https://api-docs.deepseek.com/quick_start/pricing",
|
|
35
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
36
|
+
"notes": "Current DeepSeek API model. Official docs list 1M context and 384K max output, with both non-thinking and thinking modes. DeepSeek's official docs also say deepseek-chat and deepseek-reasoner are compatibility names that will be deprecated on 2026/07/24 15:59 UTC and correspond to the non-thinking/thinking modes of this model. I did not find a current official image-input declaration, so supportsVision is left null."
|
|
37
|
+
},
|
|
38
|
+
{
|
|
39
|
+
"canonicalId": "deepseek-v4-pro",
|
|
40
|
+
"contextLength": 1000000,
|
|
41
|
+
"maxOutputTokens": 384000,
|
|
42
|
+
"reasoningSupportPolicy": "supported",
|
|
43
|
+
"supportedReasoningEfforts": [
|
|
44
|
+
"low",
|
|
45
|
+
"medium",
|
|
46
|
+
"high",
|
|
47
|
+
"xhigh"
|
|
48
|
+
],
|
|
49
|
+
"supportsVision": null,
|
|
50
|
+
"inputModalities": ["text"],
|
|
51
|
+
"outputModalities": ["text"],
|
|
52
|
+
"source": "https://api-docs.deepseek.com/quick_start/pricing",
|
|
53
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
54
|
+
"notes": "Current DeepSeek API model. Official docs show the same public capability envelope as deepseek-v4-flash: 1M context, 384K max output, and both non-thinking and thinking modes. I did not find a current official image-input declaration, so supportsVision is left null."
|
|
55
|
+
},
|
|
56
|
+
{
|
|
57
|
+
"canonicalId": "gpt-4.1",
|
|
58
|
+
"contextLength": 1047576,
|
|
59
|
+
"maxOutputTokens": 32768,
|
|
60
|
+
"reasoningSupportPolicy": "unsupported",
|
|
61
|
+
"supportedReasoningEfforts": [],
|
|
62
|
+
"supportsVision": true,
|
|
63
|
+
"inputModalities": ["text", "image"],
|
|
64
|
+
"outputModalities": ["text"],
|
|
65
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-4.1",
|
|
66
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
67
|
+
"notes": "OpenAI's current non-reasoning GPT-4.1 model. Official docs list text/image input, text output, a 1,047,576-token context window, and 32,768 max output tokens."
|
|
68
|
+
},
|
|
69
|
+
{
|
|
70
|
+
"canonicalId": "gpt-4.1-mini",
|
|
71
|
+
"contextLength": 1047576,
|
|
72
|
+
"maxOutputTokens": 32768,
|
|
73
|
+
"reasoningSupportPolicy": "unsupported",
|
|
74
|
+
"supportedReasoningEfforts": [],
|
|
75
|
+
"supportsVision": true,
|
|
76
|
+
"inputModalities": ["text", "image"],
|
|
77
|
+
"outputModalities": ["text"],
|
|
78
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-4.1-mini",
|
|
79
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
80
|
+
"notes": "Smaller, faster GPT-4.1 variant. Official docs list text/image input, text output, a 1,047,576-token context window, and 32,768 max output tokens."
|
|
81
|
+
},
|
|
82
|
+
{
|
|
83
|
+
"canonicalId": "gpt-4.1-nano",
|
|
84
|
+
"contextLength": 1047576,
|
|
85
|
+
"maxOutputTokens": 32768,
|
|
86
|
+
"reasoningSupportPolicy": "unsupported",
|
|
87
|
+
"supportedReasoningEfforts": [],
|
|
88
|
+
"supportsVision": true,
|
|
89
|
+
"inputModalities": ["text", "image"],
|
|
90
|
+
"outputModalities": ["text"],
|
|
91
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-4.1-nano",
|
|
92
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
93
|
+
"notes": "Fastest GPT-4.1 variant. Official docs list text/image input, text output, a 1,047,576-token context window, and 32,768 max output tokens. The page says it runs without a reasoning step."
|
|
94
|
+
},
|
|
95
|
+
{
|
|
96
|
+
"canonicalId": "gpt-4o",
|
|
97
|
+
"contextLength": 128000,
|
|
98
|
+
"maxOutputTokens": 16384,
|
|
99
|
+
"reasoningSupportPolicy": null,
|
|
100
|
+
"supportedReasoningEfforts": null,
|
|
101
|
+
"supportsVision": true,
|
|
102
|
+
"inputModalities": ["text", "image"],
|
|
103
|
+
"outputModalities": ["text"],
|
|
104
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-4o",
|
|
105
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
106
|
+
"notes": "Fast, intelligent, flexible GPT model. Official docs list text/image input, text output, a 128k context window, and 16,384 max output tokens. I did not find a page-level reasoning-effort matrix, so reasoning fields are left null."
|
|
107
|
+
},
|
|
108
|
+
{
|
|
109
|
+
"canonicalId": "gpt-4o-mini",
|
|
110
|
+
"contextLength": 128000,
|
|
111
|
+
"maxOutputTokens": 16384,
|
|
112
|
+
"reasoningSupportPolicy": null,
|
|
113
|
+
"supportedReasoningEfforts": null,
|
|
114
|
+
"supportsVision": true,
|
|
115
|
+
"inputModalities": ["text", "image"],
|
|
116
|
+
"outputModalities": ["text"],
|
|
117
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-4o-mini",
|
|
118
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
119
|
+
"notes": "Fast, affordable small GPT model. Official docs list text/image input, text output, a 128k context window, and 16,384 max output tokens. I did not find a page-level reasoning-effort matrix, so reasoning fields are left null."
|
|
120
|
+
},
|
|
121
|
+
{
|
|
122
|
+
"canonicalId": "gpt-5",
|
|
123
|
+
"contextLength": 400000,
|
|
124
|
+
"maxOutputTokens": 128000,
|
|
125
|
+
"reasoningSupportPolicy": "supported",
|
|
126
|
+
"supportedReasoningEfforts": [
|
|
127
|
+
"minimal",
|
|
128
|
+
"low",
|
|
129
|
+
"medium",
|
|
130
|
+
"high"
|
|
131
|
+
],
|
|
132
|
+
"supportsVision": true,
|
|
133
|
+
"inputModalities": ["text", "image"],
|
|
134
|
+
"outputModalities": ["text"],
|
|
135
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5",
|
|
136
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
137
|
+
"notes": "Previous GPT-5 generation flagship model. Official docs list text/image input, text output, a 400,000-token context window, 128,000 max output tokens, and reasoning.effort support for minimal, low, medium, and high."
|
|
138
|
+
},
|
|
139
|
+
{
|
|
140
|
+
"canonicalId": "gpt-5-mini",
|
|
141
|
+
"contextLength": 400000,
|
|
142
|
+
"maxOutputTokens": 128000,
|
|
143
|
+
"reasoningSupportPolicy": null,
|
|
144
|
+
"supportedReasoningEfforts": null,
|
|
145
|
+
"supportsVision": true,
|
|
146
|
+
"inputModalities": ["text", "image"],
|
|
147
|
+
"outputModalities": ["text"],
|
|
148
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5-mini",
|
|
149
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
150
|
+
"notes": "Faster, lower-cost GPT-5 variant. Official docs list text/image input, text output, a 400,000-token context window, and 128,000 max output tokens. The page surfaces reasoning token support but does not clearly publish a reasoning.effort matrix, so reasoning fields are left null."
|
|
151
|
+
},
|
|
152
|
+
{
|
|
153
|
+
"canonicalId": "gpt-5-nano",
|
|
154
|
+
"contextLength": 400000,
|
|
155
|
+
"maxOutputTokens": 128000,
|
|
156
|
+
"reasoningSupportPolicy": null,
|
|
157
|
+
"supportedReasoningEfforts": null,
|
|
158
|
+
"supportsVision": true,
|
|
159
|
+
"inputModalities": ["text", "image"],
|
|
160
|
+
"outputModalities": ["text"],
|
|
161
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5-nano",
|
|
162
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
163
|
+
"notes": "Fastest and cheapest GPT-5 variant. Official docs list text/image input, text output, a 400,000-token context window, and 128,000 max output tokens. The page surfaces reasoning token support but does not clearly publish a reasoning.effort matrix, so reasoning fields are left null."
|
|
164
|
+
},
|
|
165
|
+
{
|
|
166
|
+
"canonicalId": "gpt-5.4",
|
|
167
|
+
"contextLength": 1050000,
|
|
168
|
+
"maxOutputTokens": 128000,
|
|
169
|
+
"reasoningSupportPolicy": "supported",
|
|
170
|
+
"supportedReasoningEfforts": [
|
|
171
|
+
"none",
|
|
172
|
+
"low",
|
|
173
|
+
"medium",
|
|
174
|
+
"high",
|
|
175
|
+
"xhigh"
|
|
176
|
+
],
|
|
177
|
+
"supportsVision": true,
|
|
178
|
+
"inputModalities": ["text", "image"],
|
|
179
|
+
"outputModalities": ["text"],
|
|
180
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5.4",
|
|
181
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
182
|
+
"notes": "OpenAI's frontier model for complex professional work. Official docs list text/image input, text output, a 1,050,000-token context window, 128,000 max output tokens, and reasoning.effort support for none, low, medium, high, and xhigh."
|
|
183
|
+
},
|
|
184
|
+
{
|
|
185
|
+
"canonicalId": "gpt-5.4-pro",
|
|
186
|
+
"contextLength": 1050000,
|
|
187
|
+
"maxOutputTokens": 128000,
|
|
188
|
+
"reasoningSupportPolicy": "supported",
|
|
189
|
+
"supportedReasoningEfforts": [
|
|
190
|
+
"medium",
|
|
191
|
+
"high",
|
|
192
|
+
"xhigh"
|
|
193
|
+
],
|
|
194
|
+
"supportsVision": true,
|
|
195
|
+
"inputModalities": ["text", "image"],
|
|
196
|
+
"outputModalities": ["text"],
|
|
197
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5.4-pro",
|
|
198
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
199
|
+
"notes": "GPT-5.4 Pro is available in the Responses API only. Official docs list text/image input, text output, a 1,050,000-token context window, 128,000 max output tokens, and reasoning.effort support for medium, high, and xhigh."
|
|
200
|
+
},
|
|
201
|
+
{
|
|
202
|
+
"canonicalId": "gpt-5.4-mini",
|
|
203
|
+
"contextLength": 400000,
|
|
204
|
+
"maxOutputTokens": 128000,
|
|
205
|
+
"reasoningSupportPolicy": null,
|
|
206
|
+
"supportedReasoningEfforts": null,
|
|
207
|
+
"supportsVision": true,
|
|
208
|
+
"inputModalities": ["text", "image"],
|
|
209
|
+
"outputModalities": ["text"],
|
|
210
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5.4-mini",
|
|
211
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
212
|
+
"notes": "OpenAI's strongest mini model for coding, computer use, and subagents. Official docs list text/image input, text output, a 400,000-token context window, and 128,000 max output tokens. The page does not explicitly surface an effort matrix, so reasoning fields are left null."
|
|
213
|
+
},
|
|
214
|
+
{
|
|
215
|
+
"canonicalId": "gpt-5.4-nano",
|
|
216
|
+
"contextLength": 400000,
|
|
217
|
+
"maxOutputTokens": 128000,
|
|
218
|
+
"reasoningSupportPolicy": null,
|
|
219
|
+
"supportedReasoningEfforts": null,
|
|
220
|
+
"supportsVision": true,
|
|
221
|
+
"inputModalities": ["text", "image"],
|
|
222
|
+
"outputModalities": ["text"],
|
|
223
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5.4-nano",
|
|
224
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
225
|
+
"notes": "OpenAI's cheapest GPT-5.4-class model for simple high-volume tasks. Official docs list text/image input, text output, a 400,000-token context window, and 128,000 max output tokens. The page does not explicitly surface an effort matrix, so reasoning fields are left null."
|
|
226
|
+
},
|
|
227
|
+
{
|
|
228
|
+
"canonicalId": "gpt-5.5",
|
|
229
|
+
"contextLength": 1050000,
|
|
230
|
+
"maxOutputTokens": 128000,
|
|
231
|
+
"reasoningSupportPolicy": "supported",
|
|
232
|
+
"supportedReasoningEfforts": [
|
|
233
|
+
"none",
|
|
234
|
+
"low",
|
|
235
|
+
"medium",
|
|
236
|
+
"high",
|
|
237
|
+
"xhigh"
|
|
238
|
+
],
|
|
239
|
+
"supportsVision": true,
|
|
240
|
+
"inputModalities": ["text", "image"],
|
|
241
|
+
"outputModalities": ["text"],
|
|
242
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5.5",
|
|
243
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
244
|
+
"notes": "OpenAI's newest frontier model for coding and professional work. Official docs list text/image input, text output, a 1,050,000-token context window, 128,000 max output tokens, and reasoning.effort support for none, low, medium, high, and xhigh."
|
|
245
|
+
},
|
|
246
|
+
{
|
|
247
|
+
"canonicalId": "gpt-5.5-pro",
|
|
248
|
+
"contextLength": 1050000,
|
|
249
|
+
"maxOutputTokens": 128000,
|
|
250
|
+
"reasoningSupportPolicy": null,
|
|
251
|
+
"supportedReasoningEfforts": null,
|
|
252
|
+
"supportsVision": true,
|
|
253
|
+
"inputModalities": ["text", "image"],
|
|
254
|
+
"outputModalities": ["text"],
|
|
255
|
+
"source": "https://developers.openai.com/api/docs/models/gpt-5.5-pro",
|
|
256
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
257
|
+
"notes": "Higher-compute GPT-5.5 Pro variant. Official docs list text/image input, text output, a 1,050,000-token context window, 128,000 max output tokens, and Responses API availability. I did not find a page-level reasoning.effort matrix, so reasoning fields are left null."
|
|
258
|
+
},
|
|
259
|
+
{
|
|
260
|
+
"canonicalId": "gemini-2.5-pro",
|
|
261
|
+
"contextLength": 1048576,
|
|
262
|
+
"maxOutputTokens": 65536,
|
|
263
|
+
"reasoningSupportPolicy": "supported",
|
|
264
|
+
"supportedReasoningEfforts": [
|
|
265
|
+
"minimal",
|
|
266
|
+
"low",
|
|
267
|
+
"medium",
|
|
268
|
+
"high"
|
|
269
|
+
],
|
|
270
|
+
"supportsVision": true,
|
|
271
|
+
"inputModalities": ["text", "image", "video", "audio", "pdf"],
|
|
272
|
+
"outputModalities": ["text"],
|
|
273
|
+
"source": "https://ai.google.dev/gemini-api/docs/models/gemini-2.5-pro",
|
|
274
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
275
|
+
"notes": "Gemini 2.5 Pro is Google's current thinking model. Official docs list 1,048,576 input tokens, 65,536 output tokens, and multimodal input including PDF. Google's OpenAI compatibility guide says Gemini 2.5 Pro cannot turn thinking off, so this catalog keeps the verified minimal/low/medium/high subset."
|
|
276
|
+
},
|
|
277
|
+
{
|
|
278
|
+
"canonicalId": "gemini-2.5-flash",
|
|
279
|
+
"contextLength": 1048576,
|
|
280
|
+
"maxOutputTokens": 65536,
|
|
281
|
+
"reasoningSupportPolicy": "supported",
|
|
282
|
+
"supportedReasoningEfforts": [
|
|
283
|
+
"none",
|
|
284
|
+
"minimal",
|
|
285
|
+
"low",
|
|
286
|
+
"medium",
|
|
287
|
+
"high"
|
|
288
|
+
],
|
|
289
|
+
"supportsVision": true,
|
|
290
|
+
"inputModalities": ["text", "image", "video", "audio"],
|
|
291
|
+
"outputModalities": ["text"],
|
|
292
|
+
"source": "https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash",
|
|
293
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
294
|
+
"notes": "Gemini 2.5 Flash supports thinking and Google's OpenAI compatibility guide allows reasoning_effort none/minimal/low/medium/high for Gemini 2.5 models. Official docs list text, image, video, and audio input with 65,536 max output tokens."
|
|
295
|
+
},
|
|
296
|
+
{
|
|
297
|
+
"canonicalId": "gemini-2.5-flash-lite",
|
|
298
|
+
"contextLength": 1048576,
|
|
299
|
+
"maxOutputTokens": 65536,
|
|
300
|
+
"reasoningSupportPolicy": "supported",
|
|
301
|
+
"supportedReasoningEfforts": [
|
|
302
|
+
"none",
|
|
303
|
+
"minimal",
|
|
304
|
+
"low",
|
|
305
|
+
"medium",
|
|
306
|
+
"high"
|
|
307
|
+
],
|
|
308
|
+
"supportsVision": true,
|
|
309
|
+
"inputModalities": ["text", "image", "video", "audio", "pdf"],
|
|
310
|
+
"outputModalities": ["text"],
|
|
311
|
+
"source": "https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash-lite",
|
|
312
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
313
|
+
"notes": "Gemini 2.5 Flash-Lite is Google's most cost-efficient multimodal model. Official docs list text, image, video, audio, and PDF input, 1,048,576 input tokens, 65,536 output tokens, and thinking support. Google's OpenAI compatibility guide says reasoning_effort none disables thinking on Gemini 2.5 models."
|
|
314
|
+
},
|
|
315
|
+
{
|
|
316
|
+
"canonicalId": "gemini-3.1-flash-lite",
|
|
317
|
+
"contextLength": 1048576,
|
|
318
|
+
"maxOutputTokens": 65536,
|
|
319
|
+
"reasoningSupportPolicy": "supported",
|
|
320
|
+
"supportedReasoningEfforts": [
|
|
321
|
+
"minimal",
|
|
322
|
+
"low",
|
|
323
|
+
"medium",
|
|
324
|
+
"high"
|
|
325
|
+
],
|
|
326
|
+
"supportsVision": true,
|
|
327
|
+
"inputModalities": ["text", "image", "video", "audio", "pdf"],
|
|
328
|
+
"outputModalities": ["text"],
|
|
329
|
+
"source": "https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite",
|
|
330
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
331
|
+
"notes": "Current stable Gemini 3.1 Flash-Lite model. Official docs list text, image, video, audio, and PDF input, 1,048,576 input tokens, 65,536 output tokens, and thinking support. Google's OpenAI compatibility guide maps reasoning_effort minimal/low/medium/high to Gemini 3.1 Flash-Lite and says Gemini 3 models cannot disable thinking."
|
|
332
|
+
},
|
|
333
|
+
{
|
|
334
|
+
"canonicalId": "gemini-3.1-pro-preview",
|
|
335
|
+
"contextLength": 1048576,
|
|
336
|
+
"maxOutputTokens": 65536,
|
|
337
|
+
"reasoningSupportPolicy": "supported",
|
|
338
|
+
"supportedReasoningEfforts": [
|
|
339
|
+
"low",
|
|
340
|
+
"medium",
|
|
341
|
+
"high"
|
|
342
|
+
],
|
|
343
|
+
"supportsVision": true,
|
|
344
|
+
"inputModalities": ["text", "image", "video", "audio", "pdf"],
|
|
345
|
+
"outputModalities": ["text"],
|
|
346
|
+
"source": "https://ai.google.dev/gemini-api/docs/models/gemini-3.1-pro-preview",
|
|
347
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
348
|
+
"notes": "Gemini 3.1 Pro preview model. Official docs list text, image, video, audio, and PDF input, 1,048,576 input tokens, 65,536 output tokens, and thinking support. Google's OpenAI compatibility guide maps reasoning_effort low/medium/high to Gemini 3.1 Pro and says Gemini 3 models cannot disable thinking."
|
|
349
|
+
},
|
|
350
|
+
{
|
|
351
|
+
"canonicalId": "gemini-3.5-flash",
|
|
352
|
+
"contextLength": 1048576,
|
|
353
|
+
"maxOutputTokens": 65536,
|
|
354
|
+
"reasoningSupportPolicy": "supported",
|
|
355
|
+
"supportedReasoningEfforts": [
|
|
356
|
+
"minimal",
|
|
357
|
+
"low",
|
|
358
|
+
"medium",
|
|
359
|
+
"high"
|
|
360
|
+
],
|
|
361
|
+
"supportsVision": true,
|
|
362
|
+
"inputModalities": ["text", "image", "video", "audio", "pdf"],
|
|
363
|
+
"outputModalities": ["text"],
|
|
364
|
+
"source": "https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash",
|
|
365
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
366
|
+
"notes": "Gemini 3.5 Flash is Google's current stable Gemini 3.5 model. Official docs list text, image, video, audio, and PDF input, 1,048,576 input tokens, 65,536 output tokens, and thinking support. Google's OpenAI compatibility docs show reasoning_effort low on gemini-3.5-flash and do not allow turning thinking off on Gemini 3 models."
|
|
367
|
+
},
|
|
368
|
+
{
|
|
369
|
+
"canonicalId": "claude-fable-5",
|
|
370
|
+
"contextLength": 1000000,
|
|
371
|
+
"maxOutputTokens": 128000,
|
|
372
|
+
"reasoningSupportPolicy": "supported",
|
|
373
|
+
"supportedReasoningEfforts": [
|
|
374
|
+
"low",
|
|
375
|
+
"medium",
|
|
376
|
+
"high",
|
|
377
|
+
"xhigh"
|
|
378
|
+
],
|
|
379
|
+
"supportsVision": true,
|
|
380
|
+
"inputModalities": ["text", "image"],
|
|
381
|
+
"outputModalities": ["text"],
|
|
382
|
+
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview",
|
|
383
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
384
|
+
"notes": "Anthropic's most capable widely released model. Official docs list 1M context, 128k max output, text/image input, vision support, and always-on adaptive thinking. Anthropic's model-config docs expose low, medium, high, xhigh, and max effort on Fable 5; this catalog records the normalized subset and omits max because the schema does not include it."
|
|
385
|
+
},
|
|
386
|
+
{
|
|
387
|
+
"canonicalId": "claude-opus-4-8",
|
|
388
|
+
"contextLength": 1000000,
|
|
389
|
+
"maxOutputTokens": 128000,
|
|
390
|
+
"reasoningSupportPolicy": "supported",
|
|
391
|
+
"supportedReasoningEfforts": [
|
|
392
|
+
"low",
|
|
393
|
+
"medium",
|
|
394
|
+
"high",
|
|
395
|
+
"xhigh"
|
|
396
|
+
],
|
|
397
|
+
"supportsVision": true,
|
|
398
|
+
"inputModalities": ["text", "image"],
|
|
399
|
+
"outputModalities": ["text"],
|
|
400
|
+
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview",
|
|
401
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
402
|
+
"notes": "Anthropic's current Opus-tier model. Official docs list 1M context, 128k max output, and vision support. Anthropic docs expose low, medium, high, xhigh, and max effort on Opus 4.8; this catalog records the normalized subset and omits max because the schema does not include it."
|
|
403
|
+
},
|
|
404
|
+
{
|
|
405
|
+
"canonicalId": "claude-sonnet-4-6",
|
|
406
|
+
"contextLength": 1000000,
|
|
407
|
+
"maxOutputTokens": 64000,
|
|
408
|
+
"reasoningSupportPolicy": "supported",
|
|
409
|
+
"supportedReasoningEfforts": [
|
|
410
|
+
"low",
|
|
411
|
+
"medium",
|
|
412
|
+
"high"
|
|
413
|
+
],
|
|
414
|
+
"supportsVision": true,
|
|
415
|
+
"inputModalities": ["text", "image"],
|
|
416
|
+
"outputModalities": ["text"],
|
|
417
|
+
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview",
|
|
418
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
419
|
+
"notes": "Anthropic's current Sonnet-tier model. Official docs list 1M context, 64k max output, and vision support. Anthropic docs expose low, medium, high, and max effort on Sonnet 4.6; this catalog keeps the normalized low/medium/high subset and omits max because the schema does not include it."
|
|
420
|
+
},
|
|
421
|
+
{
|
|
422
|
+
"canonicalId": "claude-haiku-4-5",
|
|
423
|
+
"contextLength": 200000,
|
|
424
|
+
"maxOutputTokens": 64000,
|
|
425
|
+
"reasoningSupportPolicy": null,
|
|
426
|
+
"supportedReasoningEfforts": null,
|
|
427
|
+
"supportsVision": true,
|
|
428
|
+
"inputModalities": ["text", "image"],
|
|
429
|
+
"outputModalities": ["text"],
|
|
430
|
+
"source": "https://docs.anthropic.com/en/docs/about-claude/models/overview",
|
|
431
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
432
|
+
"notes": "Anthropic's current Haiku-tier model. Official docs list 200k context, 64k max output, and vision support. Anthropic docs clearly describe extended thinking support on Haiku 4.5, but I did not find a clean official effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
433
|
+
},
|
|
434
|
+
{
|
|
435
|
+
"canonicalId": "qwen3.7-max",
|
|
436
|
+
"contextLength": 1000000,
|
|
437
|
+
"maxOutputTokens": 64000,
|
|
438
|
+
"reasoningSupportPolicy": "supported",
|
|
439
|
+
"supportedReasoningEfforts": [
|
|
440
|
+
"none",
|
|
441
|
+
"minimal",
|
|
442
|
+
"low",
|
|
443
|
+
"medium",
|
|
444
|
+
"high"
|
|
445
|
+
],
|
|
446
|
+
"supportsVision": true,
|
|
447
|
+
"inputModalities": ["text", "image", "video"],
|
|
448
|
+
"outputModalities": ["text"],
|
|
449
|
+
"source": "https://help.aliyun.com/document_detail/3026903.html",
|
|
450
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
451
|
+
"notes": "Alibaba Cloud Model Studio docs list qwen3.7-max with a 1M context window and 64k max output. The OpenAI Responses API compatibility docs expose reasoning.effort values none, minimal, low, medium, and high for Qwen3.7 models. Official multimodal docs list Qwen3.7 Max snapshots with text, image, and video input, so the base alias is recorded as vision-capable."
|
|
452
|
+
},
|
|
453
|
+
{
|
|
454
|
+
"canonicalId": "qwen3.7-plus",
|
|
455
|
+
"contextLength": 1000000,
|
|
456
|
+
"maxOutputTokens": 64000,
|
|
457
|
+
"reasoningSupportPolicy": "supported",
|
|
458
|
+
"supportedReasoningEfforts": [
|
|
459
|
+
"none",
|
|
460
|
+
"minimal",
|
|
461
|
+
"low",
|
|
462
|
+
"medium",
|
|
463
|
+
"high"
|
|
464
|
+
],
|
|
465
|
+
"supportsVision": true,
|
|
466
|
+
"inputModalities": ["text", "image", "video"],
|
|
467
|
+
"outputModalities": ["text"],
|
|
468
|
+
"source": "https://help.aliyun.com/document_detail/3026903.html",
|
|
469
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
470
|
+
"notes": "Alibaba Cloud Model Studio docs list qwen3.7-plus with a 1M context window, 64k max output, and text, image, and video input with text output. The OpenAI Responses API compatibility docs expose reasoning.effort values none, minimal, low, medium, and high for Qwen3.7 models."
|
|
471
|
+
},
|
|
472
|
+
{
|
|
473
|
+
"canonicalId": "qwen3.6-flash",
|
|
474
|
+
"contextLength": 1000000,
|
|
475
|
+
"maxOutputTokens": 64000,
|
|
476
|
+
"reasoningSupportPolicy": "supported",
|
|
477
|
+
"supportedReasoningEfforts": [
|
|
478
|
+
"none",
|
|
479
|
+
"minimal",
|
|
480
|
+
"low",
|
|
481
|
+
"medium",
|
|
482
|
+
"high"
|
|
483
|
+
],
|
|
484
|
+
"supportsVision": true,
|
|
485
|
+
"inputModalities": ["text", "image", "video"],
|
|
486
|
+
"outputModalities": ["text"],
|
|
487
|
+
"source": "https://help.aliyun.com/document_detail/3026903.html",
|
|
488
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
489
|
+
"notes": "Alibaba Cloud Model Studio docs list qwen3.6-flash with a 1M context window, 64k max output, and text, image, and video input with text output. The OpenAI Responses API compatibility docs expose reasoning.effort values none, minimal, low, medium, and high for the supported Qwen thinking models."
|
|
490
|
+
},
|
|
491
|
+
{
|
|
492
|
+
"canonicalId": "doubao-seed-2.0-code",
|
|
493
|
+
"contextLength": null,
|
|
494
|
+
"maxOutputTokens": null,
|
|
495
|
+
"reasoningSupportPolicy": null,
|
|
496
|
+
"supportedReasoningEfforts": null,
|
|
497
|
+
"supportsVision": true,
|
|
498
|
+
"inputModalities": ["text", "image", "video"],
|
|
499
|
+
"outputModalities": ["text"],
|
|
500
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
501
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
502
|
+
"notes": "ByteDance Coze official model-service docs list doubao-seed-2.0-code as a multimodal model with context cache, enabling/disabling deep thinking, and adjustable thinking depth. I did not find a clear context window, max output ceiling, or schema-compatible effort subset for this exact model, so those fields are left null."
|
|
503
|
+
},
|
|
504
|
+
{
|
|
505
|
+
"canonicalId": "doubao-seed-2.0-pro",
|
|
506
|
+
"contextLength": null,
|
|
507
|
+
"maxOutputTokens": null,
|
|
508
|
+
"reasoningSupportPolicy": null,
|
|
509
|
+
"supportedReasoningEfforts": null,
|
|
510
|
+
"supportsVision": true,
|
|
511
|
+
"inputModalities": ["text", "image", "video"],
|
|
512
|
+
"outputModalities": ["text"],
|
|
513
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
514
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
515
|
+
"notes": "ByteDance Coze official model-service docs list doubao-seed-2.0-pro as a flagship multimodal model for agentic complex reasoning, multimodal understanding, long-context reasoning, structured generation, and video understanding. The same page says it supports enabling/disabling deep thinking and adjustable thinking depth, but does not publish a clear context window, max output ceiling, or schema-compatible effort subset."
|
|
516
|
+
},
|
|
517
|
+
{
|
|
518
|
+
"canonicalId": "doubao-seed-2.0-lite",
|
|
519
|
+
"contextLength": null,
|
|
520
|
+
"maxOutputTokens": null,
|
|
521
|
+
"reasoningSupportPolicy": null,
|
|
522
|
+
"supportedReasoningEfforts": null,
|
|
523
|
+
"supportsVision": true,
|
|
524
|
+
"inputModalities": ["text", "image", "video"],
|
|
525
|
+
"outputModalities": ["text"],
|
|
526
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
527
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
528
|
+
"notes": "ByteDance Coze official model-service docs list doubao-seed-2.0-lite as a multimodal model with context cache, enabling/disabling deep thinking, and adjustable thinking depth. The page says it supports long context but does not publish a clear numeric context window, max output ceiling, or schema-compatible effort subset."
|
|
529
|
+
},
|
|
530
|
+
{
|
|
531
|
+
"canonicalId": "doubao-seed-2.0-mini",
|
|
532
|
+
"contextLength": 256000,
|
|
533
|
+
"maxOutputTokens": null,
|
|
534
|
+
"reasoningSupportPolicy": "supported",
|
|
535
|
+
"supportedReasoningEfforts": [
|
|
536
|
+
"minimal",
|
|
537
|
+
"low",
|
|
538
|
+
"medium",
|
|
539
|
+
"high"
|
|
540
|
+
],
|
|
541
|
+
"supportsVision": true,
|
|
542
|
+
"inputModalities": ["text", "image", "video"],
|
|
543
|
+
"outputModalities": ["text"],
|
|
544
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
545
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
546
|
+
"notes": "ByteDance Coze official model-service docs list doubao-seed-2.0-mini as a multimodal model that supports 256k context and four thinking lengths. Because the documented four levels match minimal, low, medium, and high in adjacent Doubao entries, this catalog records those normalized efforts. I did not find a clear max output ceiling."
|
|
547
|
+
},
|
|
548
|
+
{
|
|
549
|
+
"canonicalId": "doubao-seed-1.8-251228",
|
|
550
|
+
"contextLength": null,
|
|
551
|
+
"maxOutputTokens": null,
|
|
552
|
+
"reasoningSupportPolicy": "supported",
|
|
553
|
+
"supportedReasoningEfforts": [
|
|
554
|
+
"minimal",
|
|
555
|
+
"low",
|
|
556
|
+
"medium",
|
|
557
|
+
"high"
|
|
558
|
+
],
|
|
559
|
+
"supportsVision": true,
|
|
560
|
+
"inputModalities": ["text", "image", "video"],
|
|
561
|
+
"outputModalities": ["text"],
|
|
562
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
563
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
564
|
+
"notes": "ByteDance Coze official model-service docs list doubao-Seed-1.8-251228 as a multimodal deep-thinking model and explicitly say it supports minimal, low, medium, and high modes. I did not find a clear context window or max output ceiling for this exact model."
|
|
565
|
+
},
|
|
566
|
+
{
|
|
567
|
+
"canonicalId": "doubao-seed-code",
|
|
568
|
+
"contextLength": null,
|
|
569
|
+
"maxOutputTokens": null,
|
|
570
|
+
"reasoningSupportPolicy": null,
|
|
571
|
+
"supportedReasoningEfforts": null,
|
|
572
|
+
"supportsVision": true,
|
|
573
|
+
"inputModalities": ["text", "image", "video"],
|
|
574
|
+
"outputModalities": ["text"],
|
|
575
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
576
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
577
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-Seed-Code as a multimodal model optimized for agentic programming tasks, with deep thinking enable/disable support and Responses API context cache. I did not find a clear context window, max output ceiling, or schema-compatible effort subset."
|
|
578
|
+
},
|
|
579
|
+
{
|
|
580
|
+
"canonicalId": "doubao-seed-1.6-flash-250828",
|
|
581
|
+
"contextLength": null,
|
|
582
|
+
"maxOutputTokens": null,
|
|
583
|
+
"reasoningSupportPolicy": null,
|
|
584
|
+
"supportedReasoningEfforts": null,
|
|
585
|
+
"supportsVision": true,
|
|
586
|
+
"inputModalities": ["text", "image", "video"],
|
|
587
|
+
"outputModalities": ["text"],
|
|
588
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
589
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
590
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-seed-1.6-flash-250828 as a multimodal model with deep thinking enable/disable support and context cache. The page describes it as an improved flash version but does not publish a clear context window, max output ceiling, or schema-compatible effort subset for this exact model."
|
|
591
|
+
},
|
|
592
|
+
{
|
|
593
|
+
"canonicalId": "doubao-seed-1-6-251015",
|
|
594
|
+
"contextLength": null,
|
|
595
|
+
"maxOutputTokens": null,
|
|
596
|
+
"reasoningSupportPolicy": "supported",
|
|
597
|
+
"supportedReasoningEfforts": [
|
|
598
|
+
"minimal",
|
|
599
|
+
"low",
|
|
600
|
+
"medium",
|
|
601
|
+
"high"
|
|
602
|
+
],
|
|
603
|
+
"supportsVision": true,
|
|
604
|
+
"inputModalities": ["text", "image", "video"],
|
|
605
|
+
"outputModalities": ["text"],
|
|
606
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
607
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
608
|
+
"notes": "ByteDance Coze official model-service docs list doubao-seed-1-6-251015 as a multimodal model that supports the reasoning_effort field with minimal, low, medium, and high. I did not find a clear context window or max output ceiling for this exact model."
|
|
609
|
+
},
|
|
610
|
+
{
|
|
611
|
+
"canonicalId": "doubao-seed-1.6-250615",
|
|
612
|
+
"contextLength": 256000,
|
|
613
|
+
"maxOutputTokens": 16000,
|
|
614
|
+
"reasoningSupportPolicy": null,
|
|
615
|
+
"supportedReasoningEfforts": null,
|
|
616
|
+
"supportsVision": true,
|
|
617
|
+
"inputModalities": ["text", "image", "video"],
|
|
618
|
+
"outputModalities": ["text"],
|
|
619
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
620
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
621
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-Seed-1.6-250615 as a multimodal deep-thinking model with 256k context and 16k max output. It supports enabling/disabling deep thinking or auto mode, but I did not find a clean effort-level subset that matches this catalog schema."
|
|
622
|
+
},
|
|
623
|
+
{
|
|
624
|
+
"canonicalId": "doubao-1-6-flash-250615",
|
|
625
|
+
"contextLength": 256000,
|
|
626
|
+
"maxOutputTokens": 16000,
|
|
627
|
+
"reasoningSupportPolicy": null,
|
|
628
|
+
"supportedReasoningEfforts": null,
|
|
629
|
+
"supportsVision": true,
|
|
630
|
+
"inputModalities": ["text", "image", "video"],
|
|
631
|
+
"outputModalities": ["text"],
|
|
632
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
633
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
634
|
+
"notes": "ByteDance Coze official model-service docs list doubao-1-6-flash-250615 as a fast multimodal deep-thinking model with text and vision understanding, 256k context, and 16k max output. It supports enabling/disabling deep thinking, but I did not find a clean effort-level subset that matches this catalog schema."
|
|
635
|
+
},
|
|
636
|
+
{
|
|
637
|
+
"canonicalId": "doubao-seed-1-6-vision-250815",
|
|
638
|
+
"contextLength": 256000,
|
|
639
|
+
"maxOutputTokens": 64000,
|
|
640
|
+
"reasoningSupportPolicy": "unsupported",
|
|
641
|
+
"supportedReasoningEfforts": [],
|
|
642
|
+
"supportsVision": true,
|
|
643
|
+
"inputModalities": ["text", "image", "video"],
|
|
644
|
+
"outputModalities": ["text"],
|
|
645
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
646
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
647
|
+
"notes": "ByteDance Coze official model-service docs list doubao-seed-1-6-vision-250815 as a multimodal model for video understanding, grounding, and GUI Agent scenarios, with 256k max context and 64k max output. Its advanced function list only shows context cache, so reasoning is recorded as unsupported."
|
|
648
|
+
},
|
|
649
|
+
{
|
|
650
|
+
"canonicalId": "doubao-1.5-pro-32k",
|
|
651
|
+
"contextLength": 32000,
|
|
652
|
+
"maxOutputTokens": 12000,
|
|
653
|
+
"reasoningSupportPolicy": "unsupported",
|
|
654
|
+
"supportedReasoningEfforts": [],
|
|
655
|
+
"supportsVision": false,
|
|
656
|
+
"inputModalities": ["text"],
|
|
657
|
+
"outputModalities": ["text"],
|
|
658
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
659
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
660
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-1.5-pro-32k as a text model with 32k context and 12k max output. The advanced function list shows context cache only, so reasoning is recorded as unsupported."
|
|
661
|
+
},
|
|
662
|
+
{
|
|
663
|
+
"canonicalId": "doubao-1.5-lite-32k-250115",
|
|
664
|
+
"contextLength": 32000,
|
|
665
|
+
"maxOutputTokens": 12000,
|
|
666
|
+
"reasoningSupportPolicy": "unsupported",
|
|
667
|
+
"supportedReasoningEfforts": [],
|
|
668
|
+
"supportsVision": false,
|
|
669
|
+
"inputModalities": ["text"],
|
|
670
|
+
"outputModalities": ["text"],
|
|
671
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
672
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
673
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-1.5-lite-32k/250115 as a text model with 32k context and 12k max output. The slash in the displayed model version is normalized to a hyphen here because model IDs are matched conservatively by canonicalId. The advanced function list shows context cache only, so reasoning is recorded as unsupported."
|
|
674
|
+
},
|
|
675
|
+
{
|
|
676
|
+
"canonicalId": "doubao-1.5-pro-vision-32k-250115",
|
|
677
|
+
"contextLength": 32000,
|
|
678
|
+
"maxOutputTokens": null,
|
|
679
|
+
"reasoningSupportPolicy": "unsupported",
|
|
680
|
+
"supportedReasoningEfforts": [],
|
|
681
|
+
"supportsVision": true,
|
|
682
|
+
"inputModalities": ["text", "image"],
|
|
683
|
+
"outputModalities": ["text"],
|
|
684
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
685
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
686
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-1.5-pro-vision-32k/250115 as a multimodal vision-understanding model. The version name itself indicates 32k context, but I did not find a clear max output ceiling. The advanced function list says no advanced features, so reasoning is recorded as unsupported."
|
|
687
|
+
},
|
|
688
|
+
{
|
|
689
|
+
"canonicalId": "doubao-1.5-pro-32k-character-250715",
|
|
690
|
+
"contextLength": 32000,
|
|
691
|
+
"maxOutputTokens": null,
|
|
692
|
+
"reasoningSupportPolicy": "unsupported",
|
|
693
|
+
"supportedReasoningEfforts": [],
|
|
694
|
+
"supportsVision": false,
|
|
695
|
+
"inputModalities": ["text"],
|
|
696
|
+
"outputModalities": ["text"],
|
|
697
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
698
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
699
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-1.5-pro-32k-character-250715 as a text role-play model. The version name itself indicates 32k context, but I did not find a clear max output ceiling. The advanced function list says no advanced features, so reasoning is recorded as unsupported."
|
|
700
|
+
},
|
|
701
|
+
{
|
|
702
|
+
"canonicalId": "doubao-lite",
|
|
703
|
+
"contextLength": 32000,
|
|
704
|
+
"maxOutputTokens": null,
|
|
705
|
+
"reasoningSupportPolicy": "unsupported",
|
|
706
|
+
"supportedReasoningEfforts": [],
|
|
707
|
+
"supportsVision": false,
|
|
708
|
+
"inputModalities": ["text"],
|
|
709
|
+
"outputModalities": ["text"],
|
|
710
|
+
"source": "https://docs.coze.cn/api/open/docs/guides/model_service",
|
|
711
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
712
|
+
"notes": "ByteDance Coze official model-service docs list Doubao-lite as a text model that supports 32k context inference and fine-tuning. I did not find a clear max output ceiling. The advanced function list says no advanced features, so reasoning is recorded as unsupported."
|
|
713
|
+
},
|
|
714
|
+
{
|
|
715
|
+
"canonicalId": "kimi-k2.7-code",
|
|
716
|
+
"contextLength": 256000,
|
|
717
|
+
"maxOutputTokens": null,
|
|
718
|
+
"reasoningSupportPolicy": null,
|
|
719
|
+
"supportedReasoningEfforts": null,
|
|
720
|
+
"supportsVision": true,
|
|
721
|
+
"inputModalities": ["text", "image", "video"],
|
|
722
|
+
"outputModalities": ["text"],
|
|
723
|
+
"source": "https://platform.kimi.com/docs/models.md",
|
|
724
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
725
|
+
"notes": "Moonshot's Kimi model list describes kimi-k2.7-code as the current coding model with a 256k context window. The Kimi K2.7 Code guide documents text, image, and video input and text output. The model is thinking-oriented, but I did not find a clean effort-level mapping that matches this catalog schema, so reasoning fields are left null."
|
|
726
|
+
},
|
|
727
|
+
{
|
|
728
|
+
"canonicalId": "kimi-k2.7-code-highspeed",
|
|
729
|
+
"contextLength": 256000,
|
|
730
|
+
"maxOutputTokens": null,
|
|
731
|
+
"reasoningSupportPolicy": null,
|
|
732
|
+
"supportedReasoningEfforts": null,
|
|
733
|
+
"supportsVision": true,
|
|
734
|
+
"inputModalities": ["text", "image", "video"],
|
|
735
|
+
"outputModalities": ["text"],
|
|
736
|
+
"source": "https://platform.kimi.com/docs/models.md",
|
|
737
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
738
|
+
"notes": "Moonshot's Kimi model list describes kimi-k2.7-code-highspeed as the high-speed version of Kimi K2.7 Code. It shares the 256k context family and multimodal text, image, and video input capability. No official max output ceiling or schema-compatible reasoning effort matrix was found."
|
|
739
|
+
},
|
|
740
|
+
{
|
|
741
|
+
"canonicalId": "kimi-k2.6",
|
|
742
|
+
"contextLength": 256000,
|
|
743
|
+
"maxOutputTokens": null,
|
|
744
|
+
"reasoningSupportPolicy": null,
|
|
745
|
+
"supportedReasoningEfforts": null,
|
|
746
|
+
"supportsVision": true,
|
|
747
|
+
"inputModalities": ["text", "image", "video"],
|
|
748
|
+
"outputModalities": ["text"],
|
|
749
|
+
"source": "https://platform.kimi.com/docs/models.md",
|
|
750
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
751
|
+
"notes": "Moonshot's Kimi model list describes kimi-k2.6 as a current 256k-context multimodal model. The Kimi K2.6 guide documents text, image, and video input and supports thinking and non-thinking modes, but the control does not map cleanly to this catalog's effort enum, so reasoning fields are left null."
|
|
752
|
+
},
|
|
753
|
+
{
|
|
754
|
+
"canonicalId": "glm-5.2",
|
|
755
|
+
"contextLength": 1000000,
|
|
756
|
+
"maxOutputTokens": 128000,
|
|
757
|
+
"reasoningSupportPolicy": null,
|
|
758
|
+
"supportedReasoningEfforts": null,
|
|
759
|
+
"supportsVision": false,
|
|
760
|
+
"inputModalities": ["text"],
|
|
761
|
+
"outputModalities": ["text"],
|
|
762
|
+
"source": "https://docs.bigmodel.cn/cn/guide/start/model-overview",
|
|
763
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
764
|
+
"notes": "Zhipu's official model overview lists GLM-5.2 in the text model table with 1M context and 128K max output. I did not find a clean official reasoning-effort subset for this model that matches this catalog schema, so reasoning fields are left null."
|
|
765
|
+
},
|
|
766
|
+
{
|
|
767
|
+
"canonicalId": "glm-5.1",
|
|
768
|
+
"contextLength": 200000,
|
|
769
|
+
"maxOutputTokens": 128000,
|
|
770
|
+
"reasoningSupportPolicy": null,
|
|
771
|
+
"supportedReasoningEfforts": null,
|
|
772
|
+
"supportsVision": false,
|
|
773
|
+
"inputModalities": ["text"],
|
|
774
|
+
"outputModalities": ["text"],
|
|
775
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/text/glm-5.1",
|
|
776
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
777
|
+
"notes": "Zhipu's official GLM-5.1 model page lists text input, text output, a 200K context window, and 128K max output tokens. The page documents thinking modes, but I did not find a clean effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
778
|
+
},
|
|
779
|
+
{
|
|
780
|
+
"canonicalId": "glm-5",
|
|
781
|
+
"contextLength": 200000,
|
|
782
|
+
"maxOutputTokens": 128000,
|
|
783
|
+
"reasoningSupportPolicy": null,
|
|
784
|
+
"supportedReasoningEfforts": null,
|
|
785
|
+
"supportsVision": false,
|
|
786
|
+
"inputModalities": ["text"],
|
|
787
|
+
"outputModalities": ["text"],
|
|
788
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/text/glm-5",
|
|
789
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
790
|
+
"notes": "Zhipu's official GLM-5 model page lists text input, text output, a 200K context window, and 128K max output tokens. The page documents thinking modes, but I did not find a clean effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
791
|
+
},
|
|
792
|
+
{
|
|
793
|
+
"canonicalId": "glm-5-turbo",
|
|
794
|
+
"contextLength": 200000,
|
|
795
|
+
"maxOutputTokens": 128000,
|
|
796
|
+
"reasoningSupportPolicy": null,
|
|
797
|
+
"supportedReasoningEfforts": null,
|
|
798
|
+
"supportsVision": false,
|
|
799
|
+
"inputModalities": ["text"],
|
|
800
|
+
"outputModalities": ["text"],
|
|
801
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/text/glm-5-turbo",
|
|
802
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
803
|
+
"notes": "Zhipu's official GLM-5-Turbo model page lists text input, text output, a 200K context window, and 128K max output tokens. The page documents thinking modes, but I did not find a clean effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
804
|
+
},
|
|
805
|
+
{
|
|
806
|
+
"canonicalId": "glm-4.7",
|
|
807
|
+
"contextLength": 200000,
|
|
808
|
+
"maxOutputTokens": 128000,
|
|
809
|
+
"reasoningSupportPolicy": null,
|
|
810
|
+
"supportedReasoningEfforts": null,
|
|
811
|
+
"supportsVision": false,
|
|
812
|
+
"inputModalities": ["text"],
|
|
813
|
+
"outputModalities": ["text"],
|
|
814
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/text/glm-4.7",
|
|
815
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
816
|
+
"notes": "Zhipu's official GLM-4.7 model page lists text input, text output, a 200K context window, and 128K max output tokens. The page documents thinking modes, but I did not find a clean effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
817
|
+
},
|
|
818
|
+
{
|
|
819
|
+
"canonicalId": "glm-4.7-flashx",
|
|
820
|
+
"contextLength": 200000,
|
|
821
|
+
"maxOutputTokens": 128000,
|
|
822
|
+
"reasoningSupportPolicy": null,
|
|
823
|
+
"supportedReasoningEfforts": null,
|
|
824
|
+
"supportsVision": false,
|
|
825
|
+
"inputModalities": ["text"],
|
|
826
|
+
"outputModalities": ["text"],
|
|
827
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/text/glm-4.7",
|
|
828
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
829
|
+
"notes": "Zhipu's GLM-4.7 page lists GLM-4.7-FlashX as a text-input, text-output GLM-4.7 family variant with a 200K context window and 128K max output tokens. The page documents thinking modes, but I did not find a clean effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
830
|
+
},
|
|
831
|
+
{
|
|
832
|
+
"canonicalId": "glm-4.7-flash",
|
|
833
|
+
"contextLength": 200000,
|
|
834
|
+
"maxOutputTokens": 128000,
|
|
835
|
+
"reasoningSupportPolicy": null,
|
|
836
|
+
"supportedReasoningEfforts": null,
|
|
837
|
+
"supportsVision": false,
|
|
838
|
+
"inputModalities": ["text"],
|
|
839
|
+
"outputModalities": ["text"],
|
|
840
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/free/glm-4.7-flash",
|
|
841
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
842
|
+
"notes": "Zhipu's official GLM-4.7-Flash page lists text input, text output, a 200K context window, and 128K max output tokens. The page documents thinking modes, but I did not find a clean effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
843
|
+
},
|
|
844
|
+
{
|
|
845
|
+
"canonicalId": "glm-4.6",
|
|
846
|
+
"contextLength": 200000,
|
|
847
|
+
"maxOutputTokens": 128000,
|
|
848
|
+
"reasoningSupportPolicy": null,
|
|
849
|
+
"supportedReasoningEfforts": null,
|
|
850
|
+
"supportsVision": false,
|
|
851
|
+
"inputModalities": ["text"],
|
|
852
|
+
"outputModalities": ["text"],
|
|
853
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/text/glm-4.6",
|
|
854
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
855
|
+
"notes": "Zhipu's official GLM-4.6 model page lists text input, text output, a 200K context window, and 128K max output tokens. The page documents deep thinking support, but I did not find a clean effort-level subset that matches this catalog schema, so reasoning fields are left null."
|
|
856
|
+
},
|
|
857
|
+
{
|
|
858
|
+
"canonicalId": "glm-5v-turbo",
|
|
859
|
+
"contextLength": 200000,
|
|
860
|
+
"maxOutputTokens": 128000,
|
|
861
|
+
"reasoningSupportPolicy": null,
|
|
862
|
+
"supportedReasoningEfforts": null,
|
|
863
|
+
"supportsVision": true,
|
|
864
|
+
"inputModalities": ["text", "image", "video", "file"],
|
|
865
|
+
"outputModalities": ["text"],
|
|
866
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/vlm/glm-5v-turbo",
|
|
867
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
868
|
+
"notes": "Zhipu's official GLM-5V-Turbo page lists video, image, text, and file input with text output, a 200K context window, and 128K max output tokens. It supports enabling or disabling thinking mode, but I did not find a clean effort-level subset that matches this catalog schema."
|
|
869
|
+
},
|
|
870
|
+
{
|
|
871
|
+
"canonicalId": "glm-4.6v",
|
|
872
|
+
"contextLength": 128000,
|
|
873
|
+
"maxOutputTokens": 32000,
|
|
874
|
+
"reasoningSupportPolicy": null,
|
|
875
|
+
"supportedReasoningEfforts": null,
|
|
876
|
+
"supportsVision": true,
|
|
877
|
+
"inputModalities": ["text", "image", "video", "file"],
|
|
878
|
+
"outputModalities": ["text"],
|
|
879
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/vlm/glm-4.6v",
|
|
880
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
881
|
+
"notes": "Zhipu's official model overview lists GLM-4.6V with 128K context and 32K max output, and the GLM-4.6V page lists video, image, text, and file input with text output. It supports enabling or disabling thinking mode, but I did not find a clean effort-level subset that matches this catalog schema."
|
|
882
|
+
},
|
|
883
|
+
{
|
|
884
|
+
"canonicalId": "glm-4.6v-flash",
|
|
885
|
+
"contextLength": 128000,
|
|
886
|
+
"maxOutputTokens": 32000,
|
|
887
|
+
"reasoningSupportPolicy": null,
|
|
888
|
+
"supportedReasoningEfforts": null,
|
|
889
|
+
"supportsVision": true,
|
|
890
|
+
"inputModalities": ["text", "image", "video", "file"],
|
|
891
|
+
"outputModalities": ["text"],
|
|
892
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/free/glm-4.6v-flash",
|
|
893
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
894
|
+
"notes": "Zhipu's official model overview lists GLM-4.6V-Flash with 128K context and 32K max output, and the GLM-4.6V-Flash page lists video, image, text, and file input with text output. It supports enabling or disabling thinking mode, but I did not find a clean effort-level subset that matches this catalog schema."
|
|
895
|
+
},
|
|
896
|
+
{
|
|
897
|
+
"canonicalId": "glm-4.1v-thinking",
|
|
898
|
+
"contextLength": 64000,
|
|
899
|
+
"maxOutputTokens": 16000,
|
|
900
|
+
"reasoningSupportPolicy": null,
|
|
901
|
+
"supportedReasoningEfforts": null,
|
|
902
|
+
"supportsVision": true,
|
|
903
|
+
"inputModalities": ["text", "image", "video"],
|
|
904
|
+
"outputModalities": ["text"],
|
|
905
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/vlm/glm-4.1v-thinking",
|
|
906
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
907
|
+
"notes": "Zhipu's official model overview lists GLM-4.1V-Thinking with 64K context and 16K max output, and the model page lists video, image, and text input with text output. It has built-in deep thinking, but I did not find a clean effort-level subset that matches this catalog schema."
|
|
908
|
+
},
|
|
909
|
+
{
|
|
910
|
+
"canonicalId": "glm-4.1v-thinking-flash",
|
|
911
|
+
"contextLength": 64000,
|
|
912
|
+
"maxOutputTokens": 16000,
|
|
913
|
+
"reasoningSupportPolicy": null,
|
|
914
|
+
"supportedReasoningEfforts": null,
|
|
915
|
+
"supportsVision": true,
|
|
916
|
+
"inputModalities": ["text", "image", "video"],
|
|
917
|
+
"outputModalities": ["text"],
|
|
918
|
+
"source": "https://docs.bigmodel.cn/cn/guide/models/free/glm-4.1v-thinking-flash",
|
|
919
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
920
|
+
"notes": "Zhipu's official model overview lists GLM-4.1V-Thinking-Flash with 64K context and 16K max output, and the model page describes it as a visual reasoning model for image and video understanding. It has built-in deep thinking, but I did not find a clean effort-level subset that matches this catalog schema."
|
|
921
|
+
},
|
|
922
|
+
{
|
|
923
|
+
"canonicalId": "MiniMax-M3",
|
|
924
|
+
"contextLength": 1000000,
|
|
925
|
+
"maxOutputTokens": null,
|
|
926
|
+
"reasoningSupportPolicy": null,
|
|
927
|
+
"supportedReasoningEfforts": null,
|
|
928
|
+
"supportsVision": true,
|
|
929
|
+
"inputModalities": ["text", "image", "video"],
|
|
930
|
+
"outputModalities": ["text"],
|
|
931
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
932
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
933
|
+
"notes": "MiniMax official text generation docs list MiniMax-M3 as a native multimodal 1M-context model. The same page says OpenAI-compatible Chat Completions support text, image, and video input. Thinking is available in MiniMax-compatible protocol surfaces, but I did not find a schema-compatible reasoning effort matrix, so reasoning fields are left null."
|
|
934
|
+
},
|
|
935
|
+
{
|
|
936
|
+
"canonicalId": "MiniMax-M2.7",
|
|
937
|
+
"contextLength": 204800,
|
|
938
|
+
"maxOutputTokens": null,
|
|
939
|
+
"reasoningSupportPolicy": null,
|
|
940
|
+
"supportedReasoningEfforts": null,
|
|
941
|
+
"supportsVision": null,
|
|
942
|
+
"inputModalities": ["text"],
|
|
943
|
+
"outputModalities": ["text"],
|
|
944
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
945
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
946
|
+
"notes": "MiniMax official text generation docs list MiniMax-M2.7 with a 204,800-token context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
947
|
+
},
|
|
948
|
+
{
|
|
949
|
+
"canonicalId": "MiniMax-M2.7-highspeed",
|
|
950
|
+
"contextLength": 204800,
|
|
951
|
+
"maxOutputTokens": null,
|
|
952
|
+
"reasoningSupportPolicy": null,
|
|
953
|
+
"supportedReasoningEfforts": null,
|
|
954
|
+
"supportsVision": null,
|
|
955
|
+
"inputModalities": ["text"],
|
|
956
|
+
"outputModalities": ["text"],
|
|
957
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
958
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
959
|
+
"notes": "MiniMax official text generation docs list MiniMax-M2.7-highspeed with a 204,800-token context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
960
|
+
},
|
|
961
|
+
{
|
|
962
|
+
"canonicalId": "MiniMax-M2.5",
|
|
963
|
+
"contextLength": 204800,
|
|
964
|
+
"maxOutputTokens": null,
|
|
965
|
+
"reasoningSupportPolicy": null,
|
|
966
|
+
"supportedReasoningEfforts": null,
|
|
967
|
+
"supportsVision": null,
|
|
968
|
+
"inputModalities": ["text"],
|
|
969
|
+
"outputModalities": ["text"],
|
|
970
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
971
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
972
|
+
"notes": "MiniMax official text generation docs list MiniMax-M2.5 with a 204,800-token context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
973
|
+
},
|
|
974
|
+
{
|
|
975
|
+
"canonicalId": "MiniMax-M2.5-highspeed",
|
|
976
|
+
"contextLength": 204800,
|
|
977
|
+
"maxOutputTokens": null,
|
|
978
|
+
"reasoningSupportPolicy": null,
|
|
979
|
+
"supportedReasoningEfforts": null,
|
|
980
|
+
"supportsVision": null,
|
|
981
|
+
"inputModalities": ["text"],
|
|
982
|
+
"outputModalities": ["text"],
|
|
983
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
984
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
985
|
+
"notes": "MiniMax official text generation docs list MiniMax-M2.5-highspeed with a 204,800-token context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
986
|
+
},
|
|
987
|
+
{
|
|
988
|
+
"canonicalId": "MiniMax-M2.1",
|
|
989
|
+
"contextLength": 204800,
|
|
990
|
+
"maxOutputTokens": null,
|
|
991
|
+
"reasoningSupportPolicy": null,
|
|
992
|
+
"supportedReasoningEfforts": null,
|
|
993
|
+
"supportsVision": null,
|
|
994
|
+
"inputModalities": ["text"],
|
|
995
|
+
"outputModalities": ["text"],
|
|
996
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
997
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
998
|
+
"notes": "MiniMax official text generation docs list MiniMax-M2.1 with a 204,800-token context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
999
|
+
},
|
|
1000
|
+
{
|
|
1001
|
+
"canonicalId": "MiniMax-M2.1-highspeed",
|
|
1002
|
+
"contextLength": 204800,
|
|
1003
|
+
"maxOutputTokens": null,
|
|
1004
|
+
"reasoningSupportPolicy": null,
|
|
1005
|
+
"supportedReasoningEfforts": null,
|
|
1006
|
+
"supportsVision": null,
|
|
1007
|
+
"inputModalities": ["text"],
|
|
1008
|
+
"outputModalities": ["text"],
|
|
1009
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
1010
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1011
|
+
"notes": "MiniMax official text generation docs list MiniMax-M2.1-highspeed with a 204,800-token context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
1012
|
+
},
|
|
1013
|
+
{
|
|
1014
|
+
"canonicalId": "MiniMax-M2",
|
|
1015
|
+
"contextLength": 204800,
|
|
1016
|
+
"maxOutputTokens": null,
|
|
1017
|
+
"reasoningSupportPolicy": null,
|
|
1018
|
+
"supportedReasoningEfforts": null,
|
|
1019
|
+
"supportsVision": null,
|
|
1020
|
+
"inputModalities": ["text"],
|
|
1021
|
+
"outputModalities": ["text"],
|
|
1022
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
1023
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1024
|
+
"notes": "MiniMax official text generation docs list MiniMax-M2 with a 204,800-token context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
1025
|
+
},
|
|
1026
|
+
{
|
|
1027
|
+
"canonicalId": "M2-her",
|
|
1028
|
+
"contextLength": 64000,
|
|
1029
|
+
"maxOutputTokens": null,
|
|
1030
|
+
"reasoningSupportPolicy": null,
|
|
1031
|
+
"supportedReasoningEfforts": null,
|
|
1032
|
+
"supportsVision": null,
|
|
1033
|
+
"inputModalities": ["text"],
|
|
1034
|
+
"outputModalities": ["text"],
|
|
1035
|
+
"source": "https://platform.minimaxi.com/docs/guides/text-generation.md",
|
|
1036
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1037
|
+
"notes": "MiniMax official text generation docs list M2-her as a chat model with a 64K context window. I did not find an official max output ceiling, vision declaration, or schema-compatible reasoning effort matrix for this model."
|
|
1038
|
+
},
|
|
1039
|
+
{
|
|
1040
|
+
"canonicalId": "step-3.7-flash",
|
|
1041
|
+
"contextLength": 256000,
|
|
1042
|
+
"maxOutputTokens": null,
|
|
1043
|
+
"reasoningSupportPolicy": "supported",
|
|
1044
|
+
"supportedReasoningEfforts": [
|
|
1045
|
+
"low",
|
|
1046
|
+
"medium",
|
|
1047
|
+
"high"
|
|
1048
|
+
],
|
|
1049
|
+
"supportsVision": true,
|
|
1050
|
+
"inputModalities": ["text", "image", "video"],
|
|
1051
|
+
"outputModalities": ["text"],
|
|
1052
|
+
"source": "https://platform.stepfun.com/docs/zh/guides/models/step-3.7-flash",
|
|
1053
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1054
|
+
"notes": "StepFun official docs describe step-3.7-flash as a 256K-context native multimodal reasoning model with image and video understanding. The same page documents reasoning_effort levels low, medium, and high for Chat Completions, plus output_config.effort for the Messages API. I did not find a clear max output ceiling."
|
|
1055
|
+
},
|
|
1056
|
+
{
|
|
1057
|
+
"canonicalId": "step-3.5-flash",
|
|
1058
|
+
"contextLength": 256000,
|
|
1059
|
+
"maxOutputTokens": null,
|
|
1060
|
+
"reasoningSupportPolicy": null,
|
|
1061
|
+
"supportedReasoningEfforts": null,
|
|
1062
|
+
"supportsVision": false,
|
|
1063
|
+
"inputModalities": ["text"],
|
|
1064
|
+
"outputModalities": ["text"],
|
|
1065
|
+
"source": "https://platform.stepfun.com/docs/zh/guides/models/step-3.5-flash",
|
|
1066
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1067
|
+
"notes": "StepFun official docs describe step-3.5-flash as a 256K-context pure text reasoning model. Variant-specific pages expose some reasoning strengths, but I did not find a clean base-model effort subset for this canonical ID, so reasoning fields are left null."
|
|
1068
|
+
},
|
|
1069
|
+
{
|
|
1070
|
+
"canonicalId": "grok-4.3",
|
|
1071
|
+
"contextLength": 1000000,
|
|
1072
|
+
"maxOutputTokens": null,
|
|
1073
|
+
"reasoningSupportPolicy": "supported",
|
|
1074
|
+
"supportedReasoningEfforts": [
|
|
1075
|
+
"none",
|
|
1076
|
+
"low",
|
|
1077
|
+
"medium",
|
|
1078
|
+
"high"
|
|
1079
|
+
],
|
|
1080
|
+
"supportsVision": true,
|
|
1081
|
+
"inputModalities": ["text", "image"],
|
|
1082
|
+
"outputModalities": ["text"],
|
|
1083
|
+
"source": "https://docs.x.ai/developers/models/grok-4.3",
|
|
1084
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1085
|
+
"notes": "xAI official docs list grok-4.3 with text and image input, text output, and a 1,000,000-token max prompt length. The model page describes configurable reasoning and links effort levels none, low, medium, and high. I did not find a clear max output ceiling."
|
|
1086
|
+
},
|
|
1087
|
+
{
|
|
1088
|
+
"canonicalId": "grok-4.20",
|
|
1089
|
+
"contextLength": 1000000,
|
|
1090
|
+
"maxOutputTokens": null,
|
|
1091
|
+
"reasoningSupportPolicy": null,
|
|
1092
|
+
"supportedReasoningEfforts": null,
|
|
1093
|
+
"supportsVision": true,
|
|
1094
|
+
"inputModalities": ["text", "image"],
|
|
1095
|
+
"outputModalities": ["text"],
|
|
1096
|
+
"source": "https://docs.x.ai/developers/models/grok-4.20-reasoning",
|
|
1097
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1098
|
+
"notes": "xAI official docs list grok-4.20 as an alias for model name grok-4.20-0309-reasoning, with text and image input, text output, a 1,000,000-token context window, and reasoning capability. I did not find an official reasoning.effort matrix for this reasoning-model alias, so reasoning fields are left null."
|
|
1099
|
+
},
|
|
1100
|
+
{
|
|
1101
|
+
"canonicalId": "grok-4.20-0309-reasoning",
|
|
1102
|
+
"contextLength": 1000000,
|
|
1103
|
+
"maxOutputTokens": null,
|
|
1104
|
+
"reasoningSupportPolicy": null,
|
|
1105
|
+
"supportedReasoningEfforts": null,
|
|
1106
|
+
"supportsVision": true,
|
|
1107
|
+
"inputModalities": ["text", "image"],
|
|
1108
|
+
"outputModalities": ["text"],
|
|
1109
|
+
"source": "https://docs.x.ai/developers/models/grok-4.20-reasoning",
|
|
1110
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1111
|
+
"notes": "xAI official docs list grok-4.20-0309-reasoning with text and image input, text output, a 1,000,000-token context window, and reasoning capability. I did not find an official reasoning.effort matrix for this exact model name, so reasoning fields are left null."
|
|
1112
|
+
},
|
|
1113
|
+
{
|
|
1114
|
+
"canonicalId": "grok-4.20-non-reasoning",
|
|
1115
|
+
"contextLength": 1000000,
|
|
1116
|
+
"maxOutputTokens": null,
|
|
1117
|
+
"reasoningSupportPolicy": "unsupported",
|
|
1118
|
+
"supportedReasoningEfforts": [],
|
|
1119
|
+
"supportsVision": true,
|
|
1120
|
+
"inputModalities": ["text", "image"],
|
|
1121
|
+
"outputModalities": ["text"],
|
|
1122
|
+
"source": "https://docs.x.ai/developers/models/grok-4.20-0309-non-reasoning",
|
|
1123
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1124
|
+
"notes": "xAI official docs list grok-4.20-non-reasoning as an alias for model name grok-4.20-0309-non-reasoning, with text and image input, text output, a 1,000,000-token context window, and Reasoning: No."
|
|
1125
|
+
},
|
|
1126
|
+
{
|
|
1127
|
+
"canonicalId": "grok-4.20-0309-non-reasoning",
|
|
1128
|
+
"contextLength": 1000000,
|
|
1129
|
+
"maxOutputTokens": null,
|
|
1130
|
+
"reasoningSupportPolicy": "unsupported",
|
|
1131
|
+
"supportedReasoningEfforts": [],
|
|
1132
|
+
"supportsVision": true,
|
|
1133
|
+
"inputModalities": ["text", "image"],
|
|
1134
|
+
"outputModalities": ["text"],
|
|
1135
|
+
"source": "https://docs.x.ai/developers/models/grok-4.20-0309-non-reasoning",
|
|
1136
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1137
|
+
"notes": "xAI official docs list grok-4.20-0309-non-reasoning with text and image input, text output, a 1,000,000-token context window, and Reasoning: No."
|
|
1138
|
+
},
|
|
1139
|
+
{
|
|
1140
|
+
"canonicalId": "grok-4.20-multi-agent",
|
|
1141
|
+
"contextLength": 1000000,
|
|
1142
|
+
"maxOutputTokens": null,
|
|
1143
|
+
"reasoningSupportPolicy": "supported",
|
|
1144
|
+
"supportedReasoningEfforts": [
|
|
1145
|
+
"low",
|
|
1146
|
+
"medium",
|
|
1147
|
+
"high",
|
|
1148
|
+
"xhigh"
|
|
1149
|
+
],
|
|
1150
|
+
"supportsVision": true,
|
|
1151
|
+
"inputModalities": ["text", "image"],
|
|
1152
|
+
"outputModalities": ["text"],
|
|
1153
|
+
"source": "https://docs.x.ai/developers/models/grok-4.20-multi-agent-0309",
|
|
1154
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1155
|
+
"notes": "xAI official docs list grok-4.20-multi-agent as an alias for model name grok-4.20-multi-agent-0309, with text and image input, text output, a 1,000,000-token context window, and reasoning capability. xAI's reasoning guide says reasoning.effort values low, medium, high, and xhigh are supported for grok-4.20-multi-agent, but this controls agent count rather than reasoning depth."
|
|
1156
|
+
},
|
|
1157
|
+
{
|
|
1158
|
+
"canonicalId": "grok-4.20-multi-agent-0309",
|
|
1159
|
+
"contextLength": 1000000,
|
|
1160
|
+
"maxOutputTokens": null,
|
|
1161
|
+
"reasoningSupportPolicy": "supported",
|
|
1162
|
+
"supportedReasoningEfforts": [
|
|
1163
|
+
"low",
|
|
1164
|
+
"medium",
|
|
1165
|
+
"high",
|
|
1166
|
+
"xhigh"
|
|
1167
|
+
],
|
|
1168
|
+
"supportsVision": true,
|
|
1169
|
+
"inputModalities": ["text", "image"],
|
|
1170
|
+
"outputModalities": ["text"],
|
|
1171
|
+
"source": "https://docs.x.ai/developers/models/grok-4.20-multi-agent-0309",
|
|
1172
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1173
|
+
"notes": "xAI official docs list grok-4.20-multi-agent-0309 with text and image input, text output, a 1,000,000-token context window, and reasoning capability. xAI's reasoning guide says reasoning.effort values low, medium, high, and xhigh are supported for grok-4.20-multi-agent, but this controls agent count rather than reasoning depth."
|
|
1174
|
+
},
|
|
1175
|
+
{
|
|
1176
|
+
"canonicalId": "grok-build-0.1",
|
|
1177
|
+
"contextLength": 256000,
|
|
1178
|
+
"maxOutputTokens": null,
|
|
1179
|
+
"reasoningSupportPolicy": null,
|
|
1180
|
+
"supportedReasoningEfforts": null,
|
|
1181
|
+
"supportsVision": true,
|
|
1182
|
+
"inputModalities": ["text", "image"],
|
|
1183
|
+
"outputModalities": ["text"],
|
|
1184
|
+
"source": "https://docs.x.ai/developers/models/grok-build-0.1",
|
|
1185
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1186
|
+
"notes": "xAI official docs list grok-build-0.1 as an early-access fast coding model with text and image input, text output, a 256,000-token context window, and reasoning capability. I did not find an official reasoning.effort matrix for this model, so reasoning fields are left null."
|
|
1187
|
+
},
|
|
1188
|
+
{
|
|
1189
|
+
"canonicalId": "mistral-medium-3-5-26-04",
|
|
1190
|
+
"contextLength": 256000,
|
|
1191
|
+
"maxOutputTokens": null,
|
|
1192
|
+
"reasoningSupportPolicy": null,
|
|
1193
|
+
"supportedReasoningEfforts": null,
|
|
1194
|
+
"supportsVision": true,
|
|
1195
|
+
"inputModalities": ["text", "image"],
|
|
1196
|
+
"outputModalities": ["text"],
|
|
1197
|
+
"source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04",
|
|
1198
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1199
|
+
"notes": "Mistral official model card identifies Mistral Medium 3.5 as a frontier-class multimodal model and shows a 256k context window. I did not find a clear max output ceiling or schema-compatible reasoning effort matrix."
|
|
1200
|
+
},
|
|
1201
|
+
{
|
|
1202
|
+
"canonicalId": "mistral-large-3-25-12",
|
|
1203
|
+
"contextLength": 256000,
|
|
1204
|
+
"maxOutputTokens": null,
|
|
1205
|
+
"reasoningSupportPolicy": null,
|
|
1206
|
+
"supportedReasoningEfforts": null,
|
|
1207
|
+
"supportsVision": true,
|
|
1208
|
+
"inputModalities": ["text", "image"],
|
|
1209
|
+
"outputModalities": ["text"],
|
|
1210
|
+
"source": "https://docs.mistral.ai/models/model-cards/mistral-large-3-25-12",
|
|
1211
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1212
|
+
"notes": "Mistral official model card identifies Mistral Large 3 as a general-purpose multimodal model and shows a 256k context window. I did not find a clear max output ceiling or schema-compatible reasoning effort matrix."
|
|
1213
|
+
},
|
|
1214
|
+
{
|
|
1215
|
+
"canonicalId": "mistral-small-4-0-26-03",
|
|
1216
|
+
"contextLength": 256000,
|
|
1217
|
+
"maxOutputTokens": null,
|
|
1218
|
+
"reasoningSupportPolicy": null,
|
|
1219
|
+
"supportedReasoningEfforts": null,
|
|
1220
|
+
"supportsVision": true,
|
|
1221
|
+
"inputModalities": ["text", "image"],
|
|
1222
|
+
"outputModalities": ["text"],
|
|
1223
|
+
"source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03",
|
|
1224
|
+
"updatedAt": "2026-06-22T00:00:00.000Z",
|
|
1225
|
+
"notes": "Mistral official model card identifies Mistral Small 4.0 as a multimodal model and shows a 256k context window. I did not find a clear max output ceiling or schema-compatible reasoning effort matrix."
|
|
1226
|
+
}
|
|
1227
|
+
]
|
|
1228
|
+
}
|