File size: 21,203 Bytes
b3b66bf
 
c919a10
53c7098
6aaa79e
59c62cc
b9f9a46
684bece
2b36c3d
809397a
88bbb22
957783c
8173b7d
684bece
53c7098
c919a10
90a0c6e
793e72e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
af4bac7
0f3449e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c919a10
17829dc
0f3449e
 
 
 
 
 
17829dc
0f3449e
 
 
 
 
 
c919a10
af4bac7
81d02d1
 
f599584
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
828cfdc
 
81d02d1
828cfdc
 
81d02d1
 
2af566a
17829dc
c919a10
828cfdc
 
 
 
 
2af566a
17829dc
828cfdc
 
 
 
 
 
 
53c7098
81d02d1
 
f599584
 
 
 
 
 
 
 
828cfdc
 
81d02d1
828cfdc
 
81d02d1
 
2af566a
17829dc
53c7098
828cfdc
 
 
 
 
2af566a
17829dc
828cfdc
 
 
 
 
 
 
59c62cc
7adbeef
 
 
 
 
 
e8d64f2
7adbeef
 
 
 
 
 
2af566a
17829dc
59c62cc
7adbeef
 
 
 
 
 
 
 
 
 
 
 
 
957783c
 
 
 
14de3cc
 
 
957783c
14de3cc
 
 
957783c
14de3cc
957783c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d878ab1
 
 
 
 
 
 
 
 
 
 
 
 
 
f6a6c6d
17829dc
f6a6c6d
d878ab1
 
 
 
 
 
 
 
 
 
 
 
 
 
65cde5f
 
 
 
 
1b07906
65cde5f
684bece
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2af566a
17829dc
684bece
 
 
 
 
 
2af566a
17829dc
684bece
 
 
 
 
 
 
 
 
 
 
 
 
 
f6a6c6d
684bece
57b3cde
 
 
 
 
 
65cde5f
 
 
 
 
1b07906
65cde5f
2b36c3d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fee4d90
2b36c3d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fee4d90
2b36c3d
 
 
 
ff5756a
2b36c3d
 
 
 
 
 
 
 
 
 
 
809397a
 
 
 
1050e5a
 
809397a
1050e5a
809397a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
88bbb22
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8173b7d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1ae4543
8173b7d
 
 
 
 
 
 
1ae4543
8173b7d
 
 
 
 
 
 
 
 
 
 
 
c919a10
8c87ef1
c919a10
b3b66bf
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
import gradio as gr
import gemini_gradio
import openai_gradio
import anthropic_gradio
import sambanova_gradio
import xai_gradio
import hyperbolic_gradio
import perplexity_gradio
import mistral_gradio
import fireworks_gradio
import cerebras_gradio
import groq_gradio
import together_gradio



with gr.Blocks(fill_height=True) as demo:
    with gr.Tab("Meta Llama"):
        with gr.Row():
            llama_model = gr.Dropdown(
                choices=[
                    'Meta-Llama-3.2-1B-Instruct',   # Llama 3.2 1B
                    'Meta-Llama-3.2-3B-Instruct',   # Llama 3.2 3B
                    'Llama-3.2-11B-Vision-Instruct',  # Llama 3.2 11B
                    'Llama-3.2-90B-Vision-Instruct',  # Llama 3.2 90B
                    'Meta-Llama-3.1-8B-Instruct',    # Llama 3.1 8B
                    'Meta-Llama-3.1-70B-Instruct',   # Llama 3.1 70B
                    'Meta-Llama-3.1-405B-Instruct'   # Llama 3.1 405B
                ],
                value='Llama-3.2-90B-Vision-Instruct',  # Default to the most advanced model
                label="Select Llama Model",
                interactive=True
            )
        
        llama_interface = gr.load(
            name=llama_model.value,
            src=sambanova_gradio.registry,
            multimodal=True,
            fill_height=True
        )
        
        def update_llama_model(new_model):
            return gr.load(
                name=new_model,
                src=sambanova_gradio.registry,
                multimodal=True,
                fill_height=True
            )
        
        llama_model.change(
            fn=update_llama_model,
            inputs=[llama_model],
            outputs=[llama_interface]
        )
        
        gr.Markdown("**Note:** You need to use a SambaNova API key from [SambaNova Cloud](https://cloud.sambanova.ai/).")
    with gr.Tab("Gemini"):
        with gr.Row():
            gemini_model = gr.Dropdown(
                choices=[
                    'gemini-1.5-flash',        # Fast and versatile performance
                    'gemini-1.5-flash-8b',     # High volume, lower intelligence tasks
                    'gemini-1.5-pro',           # Complex reasoning tasks
                    'gemini-exp-1114'          # Quality improvements
                ],
                value='gemini-1.5-pro',      # Default to the most advanced model
                label="Select Gemini Model",
                interactive=True
            )
        
        gemini_interface = gr.load(
            name=gemini_model.value,
            src=gemini_gradio.registry,
            fill_height=True
        )
        
        def update_gemini_model(new_model):
            return gr.load(
                name=new_model,
                src=gemini_gradio.registry,
                fill_height=True
            )
        
        gemini_model.change(
            fn=update_gemini_model,
            inputs=[gemini_model],
            outputs=[gemini_interface]
        )
    with gr.Tab("ChatGPT"):
        with gr.Row():
            model_choice = gr.Dropdown(
                choices=[
                    'gpt-4o',                     # Most advanced model
                    'gpt-4o-2024-08-06',          # Latest snapshot
                    'gpt-4o-2024-05-13',          # Original snapshot
                    'chatgpt-4o-latest',          # Latest ChatGPT version
                    'gpt-4o-mini',                # Small model
                    'gpt-4o-mini-2024-07-18',     # Latest mini version
                    'o1-preview',                 # Reasoning model
                    'o1-preview-2024-09-12',      # Latest o1 model snapshot
                    'o1-mini',                    # Faster reasoning model
                    'o1-mini-2024-09-12',         # Latest o1-mini model snapshot
                    'gpt-4-turbo',                # Latest GPT-4 Turbo model
                    'gpt-4-turbo-2024-04-09',     # Latest GPT-4 Turbo snapshot
                    'gpt-4-turbo-preview',         # GPT-4 Turbo preview model
                    'gpt-4-0125-preview',         # GPT-4 Turbo preview model for laziness
                    'gpt-4-1106-preview',         # Improved instruction following model
                    'gpt-4',                      # Standard GPT-4 model
                    'gpt-4-0613'                  # Snapshot of GPT-4 from June 2023
                ],
                value='gpt-4o',                 # Default to the most advanced model
                label="Select Model",
                interactive=True
            )
            
        chatgpt_interface = gr.load(
            name=model_choice.value,
            src=openai_gradio.registry,
            accept_token=True,
            fill_height=True
        )
        
        def update_model(new_model):
            return gr.load(
                name=new_model,
                src=openai_gradio.registry,
                accept_token=True,
                fill_height=True
            )
        
        model_choice.change(
            fn=update_model,
            inputs=[model_choice],
            outputs=[chatgpt_interface]
        )
    with gr.Tab("Claude"):
        with gr.Row():
            claude_model = gr.Dropdown(
                choices=[
                    'claude-3-5-sonnet-20241022',  # Latest Sonnet
                    'claude-3-5-haiku-20241022',   # Latest Haiku
                    'claude-3-opus-20240229',       # Opus
                    'claude-3-sonnet-20240229',     # Previous Sonnet
                    'claude-3-haiku-20240307'       # Previous Haiku
                ],
                value='claude-3-5-sonnet-20241022',  # Default to latest Sonnet
                label="Select Model",
                interactive=True
            )
            
        claude_interface = gr.load(
            name=claude_model.value,
            src=anthropic_gradio.registry,
            accept_token=True,
            fill_height=True
        )
        
        def update_claude_model(new_model):
            return gr.load(
                name=new_model,
                src=anthropic_gradio.registry,
                accept_token=True,
                fill_height=True
            )
        
        claude_model.change(
            fn=update_claude_model,
            inputs=[claude_model],
            outputs=[claude_interface]
        )
    with gr.Tab("Grok"):
        with gr.Row():
            grok_model = gr.Dropdown(
                choices=[
                    'grok-beta',
                    'grok-vision-beta'
                ],
                value='grok-vision-beta',
                label="Select Grok Model",
                interactive=True
            )
            
        grok_interface = gr.load(
            name=grok_model.value,
            src=xai_gradio.registry,
            fill_height=True
        )
        
        def update_grok_model(new_model):
            return gr.load(
                name=new_model,
                src=xai_gradio.registry,
                fill_height=True
            )
        
        grok_model.change(
            fn=update_grok_model,
            inputs=[grok_model],
            outputs=[grok_interface]
        )
    with gr.Tab("Groq"):
        with gr.Row():
            groq_model = gr.Dropdown(
                choices=[
                    'llama3-groq-8b-8192-tool-use-preview',
                    'llama3-groq-70b-8192-tool-use-preview',
                    'llama-3.2-1b-preview',
                    'llama-3.2-3b-preview',
                    'llama-3.2-11b-text-preview',
                    'llama-3.2-90b-text-preview',
                    'mixtral-8x7b-32768'
                ],
                value='llama3-groq-70b-8192-tool-use-preview',  # Default to Groq's optimized model
                label="Select Groq Model",
                interactive=True
            )
            
        groq_interface = gr.load(
            name=groq_model.value,
            src=groq_gradio.registry,
            fill_height=True
        )
        
        def update_groq_model(new_model):
            return gr.load(
                name=new_model,
                src=groq_gradio.registry,
                fill_height=True
            )
        
        groq_model.change(
            fn=update_groq_model,
            inputs=[groq_model],
            outputs=[groq_interface]
        )
        
        gr.Markdown("""
        **Note:** You need a Groq API key to use these models. Get one at [Groq Cloud](https://console.groq.com/).
        """)
    with gr.Tab("Qwen"):
        with gr.Row():
            qwen_model = gr.Dropdown(
                choices=[
                    'Qwen/Qwen2.5-72B-Instruct',
                    'Qwen/Qwen2.5-Coder-32B-Instruct'
                ],
                value='Qwen/Qwen2.5-72B-Instruct',
                label="Select Qwen Model",
                interactive=True
            )
            
        qwen_interface = gr.load(
            name=qwen_model.value,
            src=hyperbolic_gradio.registry,
            fill_height=True
        )
        
        def update_qwen_model(new_model):
            return gr.load(
                name=new_model,
                src=hyperbolic_gradio.registry,
                fill_height=True
            )
        
        qwen_model.change(
            fn=update_qwen_model,
            inputs=[qwen_model],
            outputs=[qwen_interface]
        )
        
        gr.Markdown("""
        <div>
            <img src="https://storage.googleapis.com/public-arena-asset/hyperbolic_logo.png" alt="Hyperbolic Logo" style="height: 50px; margin-right: 10px;">
        </div>    
                    
        **Note:** This model is supported by Hyperbolic. Build your AI apps at [Hyperbolic](https://app.hyperbolic.xyz/).
        """)
    with gr.Tab("Perplexity"):
        with gr.Row():
            perplexity_model = gr.Dropdown(
                choices=[
                    # Sonar Models (Online)
                    'llama-3.1-sonar-small-128k-online',    # 8B params
                    'llama-3.1-sonar-large-128k-online',    # 70B params
                    'llama-3.1-sonar-huge-128k-online',     # 405B params
                    # Sonar Models (Chat)
                    'llama-3.1-sonar-small-128k-chat',      # 8B params
                    'llama-3.1-sonar-large-128k-chat',      # 70B params
                    # Open Source Models
                    'llama-3.1-8b-instruct',                # 8B params
                    'llama-3.1-70b-instruct'                # 70B params
                ],
                value='llama-3.1-sonar-large-128k-online',  # Default to large online model
                label="Select Perplexity Model",
                interactive=True
            )
        
        perplexity_interface = gr.load(
            name=perplexity_model.value,
            src=perplexity_gradio.registry,
            accept_token=True,
            fill_height=True
        )
        
        def update_perplexity_model(new_model):
            return gr.load(
                name=new_model,
                src=perplexity_gradio.registry,
                accept_token=True,
                fill_height=True
            )
        
        perplexity_model.change(
            fn=update_perplexity_model,
            inputs=[perplexity_model],
            outputs=[perplexity_interface]
        )
        
        gr.Markdown("""
        **Note:** Models are grouped into three categories:
        - **Sonar Online Models**: Include search capabilities (beta access required)
        - **Sonar Chat Models**: Standard chat models
        - **Open Source Models**: Based on Hugging Face implementations
        
        For access to Online LLMs features, please fill out the [beta access form](https://perplexity.typeform.com/apiaccessform?typeform-source=docs.perplexity.ai).
        """)
    with gr.Tab("DeepSeek-V2.5"):
        gr.load(
            name='deepseek-ai/DeepSeek-V2.5',
            src=hyperbolic_gradio.registry,
            fill_height=True
        )
        gr.Markdown("""
        <div>
            <img src="https://storage.googleapis.com/public-arena-asset/hyperbolic_logo.png" alt="Hyperbolic Logo" style="height: 50px; margin-right: 10px;">
        </div>    
                    
        **Note:** This model is supported by Hyperbolic. Build your AI apps at [Hyperbolic](https://app.hyperbolic.xyz/).
        """)
    with gr.Tab("Mistral"):
        with gr.Row():
            mistral_model = gr.Dropdown(
                choices=[
                    # Premier Models
                    'mistral-large-latest',          # Top-tier reasoning model (128k)
                    'pixtral-large-latest',          # Frontier-class multimodal model (128k)
                    'ministral-3b-latest',           # Best edge model (128k)
                    'ministral-8b-latest',           # High performance edge model (128k)
                    'mistral-small-latest',          # Enterprise-grade small model (32k)
                    'codestral-latest',              # Code-specialized model (32k)
                    'mistral-embed',                 # Semantic text representation (8k)
                    'mistral-moderation-latest',     # Content moderation service (8k)
                    # Free Models
                    'pixtral-12b-2409',             # Free 12B multimodal model (128k)
                    'open-mistral-nemo',             # Multilingual model (128k)
                    'open-codestral-mamba'           # Mamba-based coding model (256k)
                ],
                value='pixtral-large-latest',    # pixtral for vision
                label="Select Mistral Model",
                interactive=True
            )
            
        mistral_interface = gr.load(
            name=mistral_model.value,
            src=mistral_gradio.registry,
            fill_height=True
        )
        
        def update_mistral_model(new_model):
            return gr.load(
                name=new_model,
                src=mistral_gradio.registry,
                fill_height=True
            )
        
        mistral_model.change(
            fn=update_mistral_model,
            inputs=[mistral_model],
            outputs=[mistral_interface],
        )
        
        gr.Markdown("""
        **Note:** You need a Mistral API key to use these models. Get one at [Mistral AI Platform](https://console.mistral.ai/).
        
        Models are grouped into two categories:
        - **Premier Models**: Require a paid API key
        - **Free Models**: Available with free API keys
        
        Each model has different context window sizes (from 8k to 256k tokens) and specialized capabilities.
        """)
    with gr.Tab("Fireworks"):
        with gr.Row():
            fireworks_model = gr.Dropdown(
                choices=[
                    'f1-preview',              # Latest F1 preview model
                    'f1-mini-preview',         # Smaller, faster model
                ],
                value='f1-preview',            # Default to preview model
                label="Select Fireworks Model",
                interactive=True
            )
            
        fireworks_interface = gr.load(
            name=fireworks_model.value,
            src=fireworks_gradio.registry,
            fill_height=True
        )
        
        def update_fireworks_model(new_model):
            return gr.load(
                name=new_model,
                src=fireworks_gradio.registry,
                fill_height=True
            )
        
        fireworks_model.change(
            fn=update_fireworks_model,
            inputs=[fireworks_model],
            outputs=[fireworks_interface]
        )
        
        gr.Markdown("""
        **Note:** You need a Fireworks AI API key to use these models. Get one at [Fireworks AI](https://app.fireworks.ai/).
        """)
    with gr.Tab("Cerebras"):
        with gr.Row():
            cerebras_model = gr.Dropdown(
                choices=[
                    'llama3.1-8b',
                    'llama3.1-70b',
                    'llama3.1-405b'
                ],
                value='llama3.1-70b',  # Default to mid-size model
                label="Select Cerebras Model",
                interactive=True
            )
            
        cerebras_interface = gr.load(
            name=cerebras_model.value,
            src=cerebras_gradio.registry,
            accept_token=True,  # Added token acceptance
            fill_height=True
        )
        
        def update_cerebras_model(new_model):
            return gr.load(
                name=new_model,
                src=cerebras_gradio.registry,
                accept_token=True,  # Added token acceptance
                fill_height=True
            )
    with gr.Tab("Together"):
        with gr.Row():
            together_model = gr.Dropdown(
                choices=[
                    # Vision Models
                    'meta-llama/Llama-Vision-Free',                     # 131k context (Free)
                    'meta-llama/Llama-3.2-11B-Vision-Instruct-Turbo',  # 131k context
                    'meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo',  # 131k context
                    # Meta Llama 3.x Models
                    'meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo',      # 131k context
                    'meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo',     # 131k context
                    'meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo',    # 130k context
                    'meta-llama/Meta-Llama-3-8B-Instruct-Turbo',        # 8k context
                    'meta-llama/Meta-Llama-3-70B-Instruct-Turbo',       # 8k context
                    'meta-llama/Llama-3.2-3B-Instruct-Turbo',          # 131k context
                    'meta-llama/Meta-Llama-3-8B-Instruct-Lite',         # 8k context, INT4
                    'meta-llama/Meta-Llama-3-70B-Instruct-Lite',        # 8k context, INT4
                    'meta-llama/Llama-3-8b-chat-hf',                    # 8k context
                    'meta-llama/Llama-3-70b-chat-hf',                   # 8k context
                    # Other Large Models
                    'nvidia/Llama-3.1-Nemotron-70B-Instruct-HF',        # 32k context
                    'Qwen/Qwen2.5-Coder-32B-Instruct',                  # 32k context
                    'microsoft/WizardLM-2-8x22B',                       # 65k context
                    'google/gemma-2-27b-it',                            # 8k context
                    'google/gemma-2-9b-it',                             # 8k context
                    'databricks/dbrx-instruct',                         # 32k context
                    # Mixtral Models
                    'mistralai/Mixtral-8x7B-Instruct-v0.1',            # 32k context
                    'mistralai/Mixtral-8x22B-Instruct-v0.1',           # 65k context
                    # Qwen Models
                    'Qwen/Qwen2.5-7B-Instruct-Turbo',                  # 32k context
                    'Qwen/Qwen2.5-72B-Instruct-Turbo',                 # 32k context
                    'Qwen/Qwen2-72B-Instruct',                         # 32k context
                    # Other Models
                    'deepseek-ai/deepseek-llm-67b-chat',               # 4k context
                    'google/gemma-2b-it',                              # 8k context
                    'Gryphe/MythoMax-L2-13b',                          # 4k context
                    'meta-llama/Llama-2-13b-chat-hf',                  # 4k context
                    'mistralai/Mistral-7B-Instruct-v0.1',              # 8k context
                    'mistralai/Mistral-7B-Instruct-v0.2',              # 32k context
                    'mistralai/Mistral-7B-Instruct-v0.3',              # 32k context
                    'NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO',     # 32k context
                    'togethercomputer/StripedHyena-Nous-7B',           # 32k context
                    'upstage/SOLAR-10.7B-Instruct-v1.0'                # 4k context
                ],
                value='meta-llama/Llama-3.2-11B-Vision-Instruct-Turbo',  # Default to recommended vision model
                label="Select Together Model",
                interactive=True
            )
            
        together_interface = gr.load(
            name=together_model.value,
            src=together_gradio.registry,
            multimodal=True,
            fill_height=True
        )
        
        def update_together_model(new_model):
            return gr.load(
                name=new_model,
                src=together_gradio.registry,
                multimodal=True,
                fill_height=True
            )
        
        together_model.change(
            fn=update_together_model,
            inputs=[together_model],
            outputs=[together_interface]
        )
        
        gr.Markdown("""
        **Note:** You need a Together AI API key to use these models. Get one at [Together AI](https://www.together.ai/).
        """)

demo.launch(ssr_mode=False)