@juspay/neurolink 8.9.0 → 8.10.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -21,6 +21,10 @@ export class MultimodalLogger {
21
21
  */
22
22
  const VISION_CAPABILITIES = {
23
23
  openai: [
24
+ // GPT-5.2 family (released Dec 11, 2025) - Latest flagship models
25
+ "gpt-5.2",
26
+ "gpt-5.2-chat-latest",
27
+ "gpt-5.2-pro",
24
28
  // GPT-5 family (released Aug 2025)
25
29
  "gpt-5",
26
30
  "gpt-5-2025-08-07",
@@ -34,6 +38,7 @@ const VISION_CAPABILITIES = {
34
38
  // o-series reasoning models (released Apr 2025)
35
39
  "o3",
36
40
  "o3-mini",
41
+ "o3-pro",
37
42
  "o4",
38
43
  "o4-mini",
39
44
  "o4-mini-deep-research",
@@ -48,12 +53,17 @@ const VISION_CAPABILITIES = {
48
53
  "gemini-3-pro-preview",
49
54
  "gemini-3-pro-preview-11-2025",
50
55
  "gemini-3-pro-latest",
56
+ "gemini-3-pro-image-preview",
51
57
  // Gemini 2.5 Series
52
58
  "gemini-2.5-pro",
53
59
  "gemini-2.5-flash",
60
+ "gemini-2.5-flash-lite",
61
+ "gemini-2.5-flash-image",
54
62
  // Gemini 2.0 Series
55
63
  "gemini-2.0-flash",
64
+ "gemini-2.0-flash-001",
56
65
  "gemini-2.0-flash-lite",
66
+ "gemini-2.0-flash-preview-image-generation",
57
67
  // Gemini 1.5 Series (Legacy)
58
68
  "gemini-1.5-pro",
59
69
  "gemini-1.5-flash",
@@ -67,24 +77,44 @@ const VISION_CAPABILITIES = {
67
77
  "claude-opus-4-5-20251101",
68
78
  "claude-haiku-4-5",
69
79
  "claude-haiku-4-5-20251001",
80
+ // Claude 4.1 and 4.0 Series
81
+ "claude-opus-4-1",
82
+ "claude-opus-4-1-20250805",
83
+ "claude-opus-4",
84
+ "claude-opus-4-20250514",
85
+ "claude-sonnet-4",
86
+ "claude-sonnet-4-20250514",
70
87
  // Claude 3.7 Series
71
88
  "claude-3-7-sonnet",
89
+ "claude-3-7-sonnet-20250219",
72
90
  // Claude 3.5 Series
73
91
  "claude-3-5-sonnet",
92
+ "claude-3-5-sonnet-20241022",
74
93
  // Claude 3 Series
75
94
  "claude-3-opus",
76
95
  "claude-3-sonnet",
77
96
  "claude-3-haiku",
78
97
  ],
79
98
  azure: [
99
+ // GPT-5.1 family (December 2025)
100
+ "gpt-5.1",
101
+ "gpt-5.1-chat",
102
+ "gpt-5.1-codex",
80
103
  // GPT-5 family
81
104
  "gpt-5",
82
105
  "gpt-5-pro",
106
+ "gpt-5-turbo",
107
+ "gpt-5-chat",
83
108
  "gpt-5-mini",
84
109
  // GPT-4.1 family
85
110
  "gpt-4.1",
86
111
  "gpt-4.1-mini",
87
112
  "gpt-4.1-nano",
113
+ // O-series
114
+ "o3",
115
+ "o3-mini",
116
+ "o3-pro",
117
+ "o4-mini",
88
118
  // Existing GPT-4
89
119
  "gpt-4o",
90
120
  "gpt-4o-mini",
@@ -97,10 +127,12 @@ const VISION_CAPABILITIES = {
97
127
  "gemini-3-pro-preview-11-2025",
98
128
  "gemini-3-pro-latest",
99
129
  "gemini-3-pro-preview",
130
+ "gemini-3-pro",
100
131
  // Gemini 2.5 models on Vertex AI
101
132
  "gemini-2.5-pro",
102
133
  "gemini-2.5-flash",
103
134
  "gemini-2.5-flash-lite",
135
+ "gemini-2.5-flash-image",
104
136
  // Gemini 2.0 models on Vertex AI
105
137
  "gemini-2.0-flash-001",
106
138
  "gemini-2.0-flash-lite",
@@ -146,27 +178,41 @@ const VISION_CAPABILITIES = {
146
178
  litellm: [
147
179
  // LiteLLM proxies to underlying providers
148
180
  // List models that support vision when going through the proxy
149
- // Gemini models
150
- "gemini-3-pro-preview",
151
- "gemini-3-pro-latest",
152
- "gemini-2.5-pro",
153
- "gemini-2.5-flash",
154
- "gemini-2.0-flash-lite",
155
- // Claude 4.5 models
181
+ // OpenAI models via LiteLLM
182
+ "openai/gpt-5",
183
+ "openai/gpt-4o",
184
+ "openai/gpt-4o-mini",
185
+ "openai/gpt-4-turbo",
186
+ "gpt-5",
187
+ "gpt-4o",
188
+ "gpt-4.1",
189
+ // Anthropic models via LiteLLM
190
+ "anthropic/claude-sonnet-4-5-20250929",
191
+ "anthropic/claude-opus-4-1-20250805",
192
+ "anthropic/claude-3-5-sonnet-20240620",
156
193
  "claude-sonnet-4-5",
157
194
  "claude-sonnet-4-5-20250929",
158
195
  "claude-opus-4-5",
159
196
  "claude-opus-4-5-20251101",
160
197
  "claude-haiku-4-5-20251001",
161
- // Claude 4 models
162
198
  "claude-sonnet-4",
163
199
  "claude-opus-4-1",
164
- // OpenAI models
165
- "gpt-4o",
166
- "gpt-4.1",
167
- "gpt-5",
200
+ // Gemini models via LiteLLM
201
+ "vertex_ai/gemini-2.5-pro",
202
+ "gemini/gemini-2.5-pro",
203
+ "gemini/gemini-2.0-flash",
204
+ "gemini-3-pro-preview",
205
+ "gemini-3-pro-latest",
206
+ "gemini-2.5-pro",
207
+ "gemini-2.5-flash",
208
+ "gemini-2.0-flash-lite",
209
+ // Groq models via LiteLLM (vision)
210
+ "groq/llama-3.2-11b-vision-preview",
168
211
  ],
169
212
  mistral: [
213
+ // Mistral Large (latest has vision via Pixtral integration)
214
+ "mistral-large-latest",
215
+ "mistral-large-2512",
170
216
  // Mistral Small 3.2 (vision support for images: PNG, JPEG, WEBP, GIF)
171
217
  "mistral-small",
172
218
  "mistral-small-latest",
@@ -176,26 +222,39 @@ const VISION_CAPABILITIES = {
176
222
  "mistral-medium",
177
223
  "mistral-medium-latest",
178
224
  "mistral-medium-3.1",
225
+ "mistral-medium-2508",
179
226
  // Magistral models (vision support)
180
227
  "magistral-small",
228
+ "magistral-small-latest",
181
229
  "magistral-medium",
230
+ "magistral-medium-latest",
182
231
  // Pixtral models (specialized vision models)
183
232
  "pixtral-12b",
184
233
  "pixtral-12b-latest",
185
234
  "pixtral-large",
186
235
  "pixtral-large-latest",
236
+ "pixtral-large-2502",
187
237
  ],
188
238
  ollama: [
189
239
  // Llama 4 family (May 2025 - Best vision + tool calling)
190
240
  "llama4:scout",
191
241
  "llama4:maverick",
192
- // Llama 3.2 vision
242
+ "llama4:latest",
243
+ "llama4",
244
+ // Llama 3.2 vision variants
193
245
  "llama3.2-vision",
246
+ "llama3.2-vision:11b",
247
+ "llama3.2-vision:90b",
194
248
  // Gemma 3 family (SigLIP vision encoder - supports tool calling + vision)
249
+ "gemma3",
195
250
  "gemma3:4b",
196
251
  "gemma3:12b",
197
252
  "gemma3:27b",
198
253
  "gemma3:latest",
254
+ // Qwen 2.5 VL (Vision-Language)
255
+ "qwen2.5-vl",
256
+ "qwen2.5-vl:72b",
257
+ "qwen2.5-vl:32b",
199
258
  // Mistral Small family (vision + tool calling)
200
259
  "mistral-small3.1",
201
260
  "mistral-small3.1:large",
@@ -203,8 +262,24 @@ const VISION_CAPABILITIES = {
203
262
  "mistral-small3.1:small",
204
263
  // LLaVA (vision-focused)
205
264
  "llava",
265
+ "llava:7b",
266
+ "llava:13b",
267
+ "llava:34b",
268
+ "llava-llama3",
269
+ "llava-llama3:8b",
206
270
  ],
207
271
  bedrock: [
272
+ // Amazon Nova models (December 2024+) - multimodal vision support
273
+ "amazon.nova-premier",
274
+ "amazon.nova-premier-v1:0",
275
+ "amazon.nova-pro",
276
+ "amazon.nova-pro-v1:0",
277
+ "amazon.nova-lite",
278
+ "amazon.nova-lite-v1:0",
279
+ "amazon.nova-2-lite-v1:0",
280
+ "nova-premier",
281
+ "nova-pro",
282
+ "nova-lite",
208
283
  // Claude 4.5 family (supports vision, PDFs, images - September-November 2025)
209
284
  "claude-sonnet-4-5",
210
285
  "claude-sonnet-4.5",
@@ -213,7 +288,7 @@ const VISION_CAPABILITIES = {
213
288
  "claude-opus-4-5",
214
289
  "claude-opus-4.5",
215
290
  "anthropic.claude-opus-4-5",
216
- "anthropic.claude-opus-4-5-20251101-v1:0",
291
+ "anthropic.claude-opus-4-5-20251124-v1:0",
217
292
  "claude-haiku-4-5",
218
293
  "claude-haiku-4.5",
219
294
  "anthropic.claude-haiku-4-5",
@@ -222,18 +297,22 @@ const VISION_CAPABILITIES = {
222
297
  "claude-sonnet-4",
223
298
  "claude-sonnet-4@",
224
299
  "anthropic.claude-sonnet-4",
300
+ "anthropic.claude-sonnet-4-20250514-v1:0",
225
301
  "claude-opus-4",
226
302
  "claude-opus-4-1",
227
303
  "claude-opus-4@",
228
304
  "anthropic.claude-opus-4",
305
+ "anthropic.claude-opus-4-1-20250805-v1:0",
229
306
  // Claude 3.7 Sonnet
230
307
  "claude-3-7-sonnet",
231
308
  "claude-3.7-sonnet",
232
309
  "anthropic.claude-3-7-sonnet",
310
+ "anthropic.claude-3-7-sonnet-20250219-v1:0",
233
311
  // Claude 3.5 Sonnet
234
312
  "claude-3-5-sonnet",
235
313
  "claude-3.5-sonnet",
236
314
  "anthropic.claude-3-5-sonnet",
315
+ "anthropic.claude-3-5-sonnet-20241022-v1:0",
237
316
  // Claude 3 Opus
238
317
  "claude-3-opus",
239
318
  "anthropic.claude-3-opus",
@@ -243,9 +322,32 @@ const VISION_CAPABILITIES = {
243
322
  // Claude 3 Haiku
244
323
  "claude-3-haiku",
245
324
  "anthropic.claude-3-haiku",
325
+ // Meta Llama 4 models (multimodal vision)
326
+ "meta.llama4-maverick-17b-instruct-v1:0",
327
+ "meta.llama4-scout-17b-instruct-v1:0",
328
+ // Meta Llama 3.2 vision models
329
+ "meta.llama3-2-90b-instruct-v1:0",
330
+ "meta.llama3-2-11b-instruct-v1:0",
331
+ // Mistral Pixtral (multimodal vision)
332
+ "mistral.pixtral-large-2502-v1:0",
246
333
  // Generic anthropic.claude prefix (catches all Claude models)
247
334
  "anthropic.claude",
248
335
  ],
336
+ huggingface: [
337
+ // Qwen 2.5 VL (Vision-Language)
338
+ "Qwen/Qwen2.5-VL-32B-Instruct",
339
+ "Qwen/Qwen2.5-VL-7B-Instruct",
340
+ // Microsoft Phi-3 Vision
341
+ "microsoft/Phi-3-vision-128k-instruct",
342
+ // LLaVA variants
343
+ "llava-hf/llava-1.5-7b-hf",
344
+ "llava-hf/llava-v1.6-mistral-7b-hf",
345
+ ],
346
+ sagemaker: [
347
+ // Meta Llama 4 vision models
348
+ "meta-llama-4-maverick-17b-128e-instruct",
349
+ "meta-llama-4-scout-17b-16e-instruct",
350
+ ],
249
351
  };
250
352
  /**
251
353
  * Provider Image Adapter - Smart routing and formatting
@@ -281,6 +383,21 @@ export class ProviderImageAdapter {
281
383
  case "ollama":
282
384
  adaptedPayload = this.formatForOpenAI(text, images);
283
385
  break;
386
+ case "huggingface":
387
+ adaptedPayload = this.formatForOpenAI(text, images);
388
+ break;
389
+ case "sagemaker":
390
+ adaptedPayload = this.formatForOpenAI(text, images);
391
+ break;
392
+ case "litellm":
393
+ adaptedPayload = this.formatForOpenAI(text, images);
394
+ break;
395
+ case "mistral":
396
+ adaptedPayload = this.formatForOpenAI(text, images);
397
+ break;
398
+ case "bedrock":
399
+ adaptedPayload = this.formatForAnthropic(text, images);
400
+ break;
284
401
  default:
285
402
  throw new Error(`Vision not supported for provider: ${provider}`);
286
403
  }