model-library 0.1.2__py3-none-any.whl → 0.1.4__py3-none-any.whl

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.
Files changed (41) hide show
  1. model_library/base/base.py +13 -6
  2. model_library/base/output.py +55 -0
  3. model_library/base/utils.py +3 -2
  4. model_library/config/README.md +169 -0
  5. model_library/config/ai21labs_models.yaml +11 -10
  6. model_library/config/alibaba_models.yaml +21 -22
  7. model_library/config/all_models.json +4708 -2471
  8. model_library/config/amazon_models.yaml +100 -102
  9. model_library/config/anthropic_models.yaml +59 -45
  10. model_library/config/cohere_models.yaml +25 -24
  11. model_library/config/deepseek_models.yaml +28 -25
  12. model_library/config/dummy_model.yaml +9 -7
  13. model_library/config/fireworks_models.yaml +86 -56
  14. model_library/config/google_models.yaml +156 -102
  15. model_library/config/inception_models.yaml +6 -6
  16. model_library/config/kimi_models.yaml +13 -14
  17. model_library/config/minimax_models.yaml +37 -0
  18. model_library/config/mistral_models.yaml +85 -29
  19. model_library/config/openai_models.yaml +192 -159
  20. model_library/config/perplexity_models.yaml +8 -23
  21. model_library/config/together_models.yaml +115 -103
  22. model_library/config/xai_models.yaml +85 -57
  23. model_library/config/zai_models.yaml +23 -15
  24. model_library/exceptions.py +12 -17
  25. model_library/file_utils.py +1 -1
  26. model_library/providers/amazon.py +32 -17
  27. model_library/providers/anthropic.py +2 -6
  28. model_library/providers/google/google.py +35 -29
  29. model_library/providers/minimax.py +33 -0
  30. model_library/providers/mistral.py +10 -1
  31. model_library/providers/openai.py +10 -8
  32. model_library/providers/together.py +18 -211
  33. model_library/register_models.py +36 -38
  34. model_library/registry_utils.py +18 -16
  35. model_library/utils.py +2 -2
  36. {model_library-0.1.2.dist-info → model_library-0.1.4.dist-info}/METADATA +3 -4
  37. model_library-0.1.4.dist-info/RECORD +64 -0
  38. model_library-0.1.2.dist-info/RECORD +0 -61
  39. {model_library-0.1.2.dist-info → model_library-0.1.4.dist-info}/WHEEL +0 -0
  40. {model_library-0.1.2.dist-info → model_library-0.1.4.dist-info}/licenses/LICENSE +0 -0
  41. {model_library-0.1.2.dist-info → model_library-0.1.4.dist-info}/top_level.txt +0 -0
model_library/config/cohere_models.yaml CHANGED
@@ -2,11 +2,12 @@ base-config:
2
2
  company: Cohere
3
3
  open_source: false
4
4
  documentation_url: https://docs.cohere.com/v2/docs/models
5
- class_properties:
5
+ supports:
6
+ images: false
7
+ files: false
8
+ tools: true
9
+ metadata:
6
10
  available_as_evaluator: false
7
- supports_images: false
8
- supports_files: false
9
- supports_tools: true
10
11
  available_for_everyone: true
11
12
  ignored_for_cost: false
12
13
  properties:
@@ -16,8 +17,8 @@ command-models:
16
17
  base-config:
17
18
  properties:
18
19
  context_window: 128000
19
- class_properties:
20
- supports_temperature: true
20
+ supports:
21
+ temperature: true
21
22
  default_parameters:
22
23
  temperature: 0.3
23
24
 
@@ -30,7 +31,7 @@ command-models:
30
31
  release_date: 2025-03-13
31
32
  properties:
32
33
  context_window: 256000
33
- max_token_output: 8000
34
+ max_tokens: 8000
34
35
  costs_per_million_token:
35
36
  input: 2.5
36
37
  output: 10.0
@@ -43,8 +44,8 @@ command-models:
43
44
  release_date: 2024-04-24
44
45
  properties:
45
46
  context_window: 4000
46
- max_token_output: 4000
47
- class_properties:
47
+ max_tokens: 4000
48
+ metadata:
48
49
  deprecated: true
49
50
  costs_per_million_token:
50
51
  input: 0.3
@@ -58,8 +59,8 @@ command-models:
58
59
  release_date: 2024-03-24
59
60
  properties:
60
61
  context_window: 4000
61
- max_token_output: 4000
62
- class_properties:
62
+ max_tokens: 4000
63
+ metadata:
63
64
  deprecated: true
64
65
  costs_per_million_token:
65
66
  input: 1.0
@@ -73,8 +74,8 @@ command-models:
73
74
  release_date: 2024-04-24
74
75
  properties:
75
76
  context_window: 128000
76
- max_token_output: 4000
77
- class_properties:
77
+ max_tokens: 4000
78
+ metadata:
78
79
  available_for_everyone: false
79
80
  deprecated: true
80
81
  costs_per_million_token:
@@ -90,8 +91,8 @@ command-models:
90
91
  release_date: 2024-03-24
91
92
  properties:
92
93
  context_window: 128000
93
- max_token_output: 4000
94
- class_properties:
94
+ max_tokens: 4000
95
+ metadata:
95
96
  available_for_everyone: false
96
97
  deprecated: true
97
98
  costs_per_million_token:
@@ -107,8 +108,8 @@ command-models:
107
108
  release_date: 2024-08-30
108
109
  properties:
109
110
  context_window: 128000
110
- max_token_output: 4000
111
- class_properties:
111
+ max_tokens: 4000
112
+ metadata:
112
113
  available_for_everyone: false
113
114
  deprecated: true
114
115
  costs_per_million_token:
@@ -124,8 +125,8 @@ command-models:
124
125
  release_date: 2024-04-24
125
126
  properties:
126
127
  context_window: 128000
127
- max_token_output: 4000
128
- class_properties:
128
+ max_tokens: 4000
129
+ metadata:
129
130
  deprecated: true
130
131
  costs_per_million_token:
131
132
  input: 2.5
@@ -140,8 +141,8 @@ command-models:
140
141
  release_date: 2024-04-24
141
142
  properties:
142
143
  context_window: 128000
143
- max_token_output: 4000
144
- class_properties:
144
+ max_tokens: 4000
145
+ metadata:
145
146
  deprecated: true
146
147
  costs_per_million_token:
147
148
  input: 2.5
@@ -155,8 +156,8 @@ command-models:
155
156
  release_date: 2024-08-30
156
157
  properties:
157
158
  context_window: 128000
158
- max_token_output: 4000
159
- class_properties:
159
+ max_tokens: 4000
160
+ metadata:
160
161
  deprecated: true
161
162
  costs_per_million_token:
162
163
  input: 2.5
@@ -171,7 +172,7 @@ command-models:
171
172
  release_date: 2024-12-13
172
173
  properties:
173
174
  context_window: 128000
174
- max_token_output: 4000
175
+ max_tokens: 4000
175
176
  costs_per_million_token:
176
177
  input: 0.0375
177
178
  output: 0.15
model_library/config/deepseek_models.yaml CHANGED
@@ -2,48 +2,51 @@ base-config:
2
2
  company: DeepSeek
3
3
  documentation_url: https://api-docs.deepseek.com/
4
4
  open_source: true
5
- class_properties:
6
- supports_images: false
5
+ supports:
6
+ images: false
7
+ files: false
8
+ metadata:
7
9
  available_as_evaluator: false
8
- supports_files: false
9
10
  available_for_everyone: true
10
11
  ignored_for_cost: false
11
12
  properties:
12
13
  reasoning_model: false
13
-
14
- deepseek-v3p2-exp-models:
14
+
15
+ deepseek-endpoints:
15
16
  base-config:
16
- class_properties:
17
- supports_images: false
18
- supports_files: false
19
- supports_batch_requests: false
20
- supports_temperature: true
17
+ supports:
18
+ images: false
19
+ files: false
20
+ tools: true
21
+ batch: false
22
+ temperature: true
21
23
  default_parameters:
22
24
  temperature: 1
23
- costs_per_million_token:
24
- input: 0.28
25
- output: 0.42
26
- cache:
27
- read_discount: 0.1
28
25
 
29
26
  deepseek/deepseek-chat:
30
- label: DeepSeek V3.2-Exp (Nonthinking)
31
- description: DeepSeek V3.2-Exp model for general conversation
27
+ label: DeepSeek Chat
28
+ description: DeepSeek's latest chat model
32
29
  release_date: 2025-09-29
33
30
  properties:
34
31
  context_window: 128_000
35
- max_token_output: 8_000
32
+ max_tokens: 8_000
36
33
  reasoning_model: false
37
- class_properties:
38
- supports_tools: true
34
+ costs_per_million_token:
35
+ input: 0.27
36
+ output: 1.10
37
+ cache:
38
+ read: 0.07
39
39
 
40
40
  deepseek/deepseek-reasoner:
41
- label: DeepSeek V3.2-Exp (Thinking)
42
- description: DeepSeek V3.2-Exp model with thinking mode for complex reasoning
41
+ label: DeepSeek Reasoner
42
+ description: DeepSeek's latest reasoning model
43
43
  release_date: 2025-09-29
44
44
  properties:
45
45
  context_window: 128_000
46
- max_token_output: 64_000
46
+ max_tokens: 64_000
47
47
  reasoning_model: true
48
- class_properties:
49
- supports_tools: false
48
+ costs_per_million_token:
49
+ input: 0.55
50
+ output: 2.19
51
+ cache:
52
+ read: 0.14
model_library/config/dummy_model.yaml CHANGED
@@ -2,11 +2,11 @@ vals-models:
2
2
  base-config:
3
3
  company: Vals AI
4
4
  documentation_url: ""
5
- class_properties:
6
- supports_batch_requests: true
5
+ supports:
6
+ batch: true
7
7
  properties:
8
8
  context_window: 128_000
9
- max_token_output: 16_384
9
+ max_tokens: 16_384
10
10
  training_cutoff: ""
11
11
 
12
12
  vals/dumbmar-5o-ultra-thinking:
@@ -14,8 +14,9 @@ vals-models:
14
14
  open_source: false
15
15
  description: Vals Dummy Model for testing
16
16
  release_date: null
17
- class_properties:
18
- supports_images: false
17
+ supports:
18
+ images: false
19
+ metadata:
19
20
  available_as_evaluator: false
20
21
  costs_per_million_token:
21
22
  input: 0.15
@@ -28,8 +29,9 @@ vals-models:
28
29
  open_source: false
29
30
  description: Vals Dummy Model for evaluating
30
31
  release_date: null
31
- class_properties:
32
- supports_images: false
32
+ supports:
33
+ images: false
34
+ metadata:
33
35
  available_as_evaluator: true
34
36
  costs_per_million_token:
35
37
  input: 0.15
model_library/config/fireworks_models.yaml CHANGED
@@ -2,12 +2,13 @@ base-config:
2
2
  company: Fireworks
3
3
  documentation_url: https://fireworks.ai/models
4
4
  open_source: false
5
- class_properties:
5
+ supports:
6
+ files: false
7
+ tools: true
8
+ metadata:
6
9
  available_as_evaluator: false
7
- supports_files: false
8
10
  available_for_everyone: true
9
11
  ignored_for_cost: false
10
- supports_tools: true
11
12
  properties:
12
13
  reasoning_model: false
13
14
  costs_per_million_token:
@@ -21,8 +22,8 @@ qwen-models:
21
22
  base-config:
22
23
  company: Alibaba
23
24
  open_source: true
24
- class_properties:
25
- supports_temperature: true
25
+ supports:
26
+ temperature: true
26
27
  default_parameters:
27
28
  temperature: 0.7
28
29
 
@@ -32,11 +33,11 @@ qwen-models:
32
33
  release_date: 2025-04-28
33
34
  properties:
34
35
  context_window: 128_000
35
- max_token_output: 32_768
36
+ max_tokens: 32_768
36
37
  training_cutoff: "2024-08"
37
38
  reasoning_model: true
38
- class_properties:
39
- supports_images: false
39
+ supports:
40
+ images: false
40
41
  costs_per_million_token:
41
42
  input: 0.22
42
43
  output: 0.88
@@ -52,13 +53,14 @@ llama-4-models:
52
53
  release_date: 2025-04-05
53
54
  properties:
54
55
  context_window: 1_000_000
55
- max_token_output: 16_384
56
+ max_tokens: 16_384
56
57
  training_cutoff: "2024-08"
57
58
  costs_per_million_token:
58
59
  input: 0.22
59
60
  output: 0.88
60
- class_properties:
61
- supports_images: true
61
+ supports:
62
+ images: true
63
+ metadata:
62
64
  deprecated: true
63
65
 
64
66
  fireworks/llama4-scout-instruct-basic:
@@ -67,23 +69,23 @@ llama-4-models:
67
69
  release_date: 2025-04-05
68
70
  properties:
69
71
  context_window: 10_000_000
70
- max_token_output: 16_384
72
+ max_tokens: 16_384
71
73
  training_cutoff: "2024-08"
72
74
  costs_per_million_token:
73
- input: 0.18
74
- output: 0.59
75
- class_properties:
76
- supports_images: true
75
+ input: 0.15
76
+ output: 0.6
77
+ supports:
78
+ images: true
79
+ metadata:
77
80
  deprecated: true
78
81
 
79
82
  deepseek-models:
80
83
  base-config:
81
84
  company: DeepSeek
82
85
  open_source: true
83
- class_properties:
84
- supports_images: false
85
- supports_temperature: true
86
- deprecated: true
86
+ supports:
87
+ images: false
88
+ temperature: true
87
89
  default_parameters:
88
90
  temperature: 1
89
91
 
@@ -93,12 +95,14 @@ deepseek-models:
93
95
  release_date: 2025-01-20
94
96
  properties:
95
97
  context_window: 163_840
96
- max_token_output: 163_840
98
+ max_tokens: 163_840
97
99
  training_cutoff: null
98
100
  reasoning_model: true
101
+ metadata:
102
+ deprecated: true
99
103
  costs_per_million_token:
100
- input: 3.00
101
- output: 8.00
104
+ input: 1.35
105
+ output: 5.4
102
106
 
103
107
  fireworks/deepseek-v3-0324:
104
108
  label: DeepSeek V3 (03/24/2025)
@@ -106,10 +110,12 @@ deepseek-models:
106
110
  release_date: 2025-03-24
107
111
  properties:
108
112
  context_window: 131_072
109
- max_token_output: 131_072
113
+ max_tokens: 131_072
114
+ metadata:
115
+ deprecated: true
110
116
  costs_per_million_token:
111
- input: 1.20
112
- output: 1.20
117
+ input: 0.9
118
+ output: 0.9
113
119
 
114
120
  fireworks/deepseek-v3:
115
121
  label: DeepSeek V3
@@ -117,10 +123,12 @@ deepseek-models:
117
123
  release_date: 2024-12-26
118
124
  properties:
119
125
  context_window: 131_072
120
- max_token_output: 131_072
126
+ max_tokens: 131_072
127
+ metadata:
128
+ deprecated: true
121
129
  costs_per_million_token:
122
- input: 0.90
123
- output: 0.90
130
+ input: 0.9
131
+ output: 0.9
124
132
 
125
133
  fireworks/deepseek-v3p1:
126
134
  label: DeepSeek V3.1
@@ -128,19 +136,39 @@ deepseek-models:
128
136
  release_date: 2025-08-21
129
137
  properties:
130
138
  context_window: 163_840
131
- max_token_output: 163_840
132
- reasoning_model: false # actually hybrid
133
- deprecated: false # still using this one
139
+ max_tokens: 163_840
140
+ reasoning_model: false
141
+ costs_per_million_token:
142
+ input: 0.56
143
+ output: 1.68
144
+
145
+ fireworks/deepseek-v3p2:
146
+ label: DeepSeek V3.2 (Nonthinking)
147
+ description: ""
148
+ release_date: 2025-12-01
149
+ properties:
150
+ context_window: 160_000
151
+ max_tokens: 20_480
152
+ reasoning_model: false
134
153
  costs_per_million_token:
135
154
  input: 0.56
136
155
  output: 1.68
156
+ cache:
157
+ read: 0.28
158
+ alternative_keys:
159
+ - fireworks/deepseek-v3p2-thinking:
160
+ label: DeepSeek V3.2 (Thinking)
161
+ properties:
162
+ reasoning_model: true
163
+ default_parameters:
164
+ reasoning_effort: "high"
137
165
 
138
166
  openai-models:
139
167
  base-config:
140
168
  company: OpenAI
141
169
  open_source: true
142
- class_properties:
143
- supports_images: false
170
+ supports:
171
+ images: false
144
172
 
145
173
  fireworks/gpt-oss-120b:
146
174
  label: GPT OSS 120B
@@ -148,7 +176,7 @@ openai-models:
148
176
  release_date: 2025-08-05
149
177
  properties:
150
178
  context_window: 128_000
151
- max_token_output: 32_768
179
+ max_tokens: 32_768
152
180
  training_cutoff: null
153
181
  reasoning_model: true
154
182
  costs_per_million_token:
@@ -161,20 +189,20 @@ openai-models:
161
189
  release_date: 2025-08-05
162
190
  properties:
163
191
  context_window: 128_000
164
- max_token_output: 32_768
192
+ max_tokens: 32_768
165
193
  training_cutoff: null
166
194
  reasoning_model: true
167
195
  costs_per_million_token:
168
- input: 0.05
169
- output: 0.20
196
+ input: 0.07
197
+ output: 0.3
170
198
 
171
199
  kimi-models:
172
200
  base-config:
173
201
  company: Kimi
174
202
  open_source: true
175
203
  documentation_url: https://www.kimi.com/
176
- class_properties:
177
- supports_images: false
204
+ supports:
205
+ images: false
178
206
 
179
207
  fireworks/kimi-k2-instruct-0905:
180
208
  label: Kimi K2 Instruct 0905
@@ -182,11 +210,11 @@ kimi-models:
182
210
  release_date: 2025-09-04
183
211
  properties:
184
212
  context_window: 256_000
185
- max_token_output: 256_000
213
+ max_tokens: 256_000
186
214
  training_cutoff: null
187
215
  reasoning_model: false
188
- class_properties:
189
- supports_images: false
216
+ supports:
217
+ images: false
190
218
  costs_per_million_token:
191
219
  input: 0.60
192
220
  output: 2.50
@@ -196,33 +224,35 @@ minimax-models:
196
224
  company: MiniMax AI
197
225
  documentation_url: https://platform.minimax.io/docs
198
226
  open_source: true
199
- class_properties:
227
+ supports:
228
+ images: false
229
+ files: false
230
+ tools: true
231
+ temperature: true
232
+ metadata:
200
233
  available_as_evaluator: false
201
- supports_images: false
202
- supports_files: false
203
- supports_tools: true
204
234
  available_for_everyone: true
205
235
  ignored_for_cost: false
206
- supports_temperature: true
207
-
236
+ default_parameters:
237
+ temperature: 1.0
238
+ top_p: 0.95
239
+ top_k: 40
208
240
 
209
241
  fireworks/minimax-m2:
210
242
  label: MiniMax-M2
211
243
  description: MiniMax-M2 is a cost-efficient open-source model optimized for agentic applications and coding in particular.
212
244
  release_date: 2025-10-26
213
245
  properties:
214
- context_window: 204_800
215
- max_token_output: 131_000
246
+ context_window: 204_800
247
+ max_tokens: 131_000
216
248
  reasoning_model: true
217
249
  training_cutoff: null
218
- class_properties:
219
- supports_tools: true
220
- supports_temperature: true
250
+ supports:
251
+ tools: true
252
+ temperature: true
221
253
  default_parameters: # taken from https://huggingface.co/MiniMaxAI/MiniMax-M2#inference-parameters
222
254
  temperature: 1.0
223
255
  top_p: 0.95
224
- top_k: 40
225
256
  costs_per_million_token:
226
257
  input: 0.30
227
258
  output: 1.20
228
-