-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathmodels.ts
More file actions
168 lines (161 loc) · 6.06 KB
/
Copy pathmodels.ts
File metadata and controls
168 lines (161 loc) · 6.06 KB
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
export interface InferenceModel {
hfModel: string;
providerModel: string;
task?: string;
// You can set this to force the value, e.g. to keep a model as 'staging' even if it's
// warm/live on Replicate. If not set, the status will be inferred from the provider model
status?: 'live' | 'staging';
}
export const inferenceModels: InferenceModel[] = [
// Text-to-Image models
{
hfModel: "black-forest-labs/FLUX.1-dev",
providerModel: "black-forest-labs/flux-dev",
},
{
hfModel: "black-forest-labs/FLUX.1-Krea-dev",
providerModel: "black-forest-labs/flux-krea-dev",
},
{
hfModel: "black-forest-labs/FLUX.1-schnell",
providerModel: "black-forest-labs/flux-schnell",
},
{
hfModel: "briaai/FIBO",
providerModel: "bria/fibo",
},
// Removed 2026-05-06: these route through pinned/community Replicate models.
// HF's Replicate adapter cannot reliably serve those because pinned requests go
// through /v1/predictions and unpinned community models do not expose deployments.
// Re-enable only if they get official deployment endpoints or HF supports pinned routes.
// - ByteDance/SDXL-Lightning -> bytedance/sdxl-lightning-4step
// - ByteDance/Hyper-SD -> bytedance/hyper-flux-16step
// - playgroundai/playground-v2.5-1024px-aesthetic -> playgroundai/playground-v2.5-1024px-aesthetic
{
hfModel: "HiDream-ai/HiDream-I1-Fast",
providerModel: "prunaai/hidream-l1-fast",
},
{
hfModel: "Qwen/Qwen-Image",
providerModel: "qwen/qwen-image",
},
{
hfModel: "Qwen/Qwen-Image-Edit",
providerModel: "qwen/qwen-image-edit",
},
// Not adding Qwen/Qwen-Image-Edit-2509 or Qwen/Qwen-Image-Edit-2511 yet:
// those Replicate schemas expect `image` as an array, while HF's current
// image-to-image adapter sends `image` as a string and `images` as an array.
// Re-test after the adapter can target array-valued image inputs.
{
hfModel: "stabilityai/stable-diffusion-3-medium",
providerModel: "stability-ai/stable-diffusion-3",
},
{
hfModel: "stabilityai/stable-diffusion-3.5-medium",
providerModel: "stability-ai/stable-diffusion-3.5-medium",
},
{
hfModel: "stabilityai/stable-diffusion-3.5-large",
providerModel: "stability-ai/stable-diffusion-3.5-large",
},
{
hfModel: "stabilityai/stable-diffusion-3.5-large-turbo",
providerModel: "stability-ai/stable-diffusion-3.5-large-turbo",
},
// Removed 2026-05-06: stability-ai/sdxl has the same pinned/community routing issue
// described above for the ByteDance and Playground mappings.
// - stabilityai/stable-diffusion-xl-base-1.0 -> stability-ai/sdxl
{
hfModel: "tencent/HunyuanImage-2.1",
providerModel: "tencent/hunyuan-image-2.1",
},
{
hfModel: "tencent/HunyuanImage-3.0",
providerModel: "tencent/hunyuan-image-3",
},
// Removed 2026-05-06: community model with no unpinned deployment endpoint.
// - zeke/rider-waite-tarot-flux -> tarot-cards/rider-waite
{
hfModel: "Tongyi-MAI/Z-Image-Turbo",
providerModel: "prunaai/z-image-turbo",
},
{
hfModel: "black-forest-labs/FLUX.2-dev",
providerModel: "black-forest-labs/flux-2-dev",
},
{
hfModel: "Qwen/Qwen-Image-2512",
providerModel: "qwen/qwen-image-2512",
},
// Text-to-Video models
// Removed 2026-05-06: Wan2.1 variants exceed Replicate's 60s sync wait limit,
// so HF receives incomplete predictions and reports malformed text-to-video output.
// - Wan-AI/Wan2.1-T2V-14B -> wavespeedai/wan-2.1-t2v-480p
// - Wan-AI/Wan2.1-T2V-1.3B -> wan-video/wan-2.1-1.3b
{
hfModel: "Wan-AI/Wan2.2-T2V-A14B",
providerModel: "wan-video/wan-2.2-t2v-fast",
task: "text-to-video",
},
{
hfModel: "Wan-AI/Wan2.2-TI2V-5B",
providerModel: "wan-video/wan-2.2-5b-fast",
task: "text-to-video",
},
{
hfModel: "Wan-AI/Wan2.2-T2V-A14B-Diffusers",
providerModel: "wan-video/wan-2.2-t2v-fast",
task: "text-to-video",
},
// Removed 2026-05-06: Veo is gated on Replicate, so HF cannot create predictions.
// - akhaliq/veo3.1-fast -> google/veo-3.1-fast
// Image-to-Video models
{
hfModel: "Wan-AI/Wan2.2-I2V-A14B",
providerModel: "wan-video/wan-2.2-i2v-fast",
},
{
hfModel: "Lightricks/LTX-Video",
providerModel: "lightricks/ltx-video",
},
// Image-to-Image models
{
hfModel: "black-forest-labs/FLUX.1-Kontext-dev",
providerModel: "black-forest-labs/flux-kontext-dev",
task: "image-to-image",
},
{
hfModel: "black-forest-labs/FLUX.2-klein-4B",
providerModel: "black-forest-labs/flux-2-klein-4b",
task: "image-to-image"
},
{
hfModel: "black-forest-labs/FLUX.2-klein-base-4B",
providerModel: "black-forest-labs/flux-2-klein-4b-base",
task: "image-to-image"
},
{
hfModel: "black-forest-labs/FLUX.2-klein-base-9B",
providerModel: "black-forest-labs/flux-2-klein-9b-base",
task: "image-to-image"
},
// Text-to-Speech models
// Removed 2026-05-06: Kokoro is a pinned community model with no HF-compatible
// deployment route. Chatterbox covers the high-impact TTS slot with a compatible schema.
// - hexgrad/Kokoro-82M -> jaaari/kokoro-82m
{
hfModel: "ResembleAI/chatterbox",
providerModel: "resemble-ai/chatterbox-turbo",
task: "text-to-speech",
},
// Automatic Speech Recognition models
// Removed 2026-05-06: cold pinned community ASR mapping; re-test with an audio
// fixture before re-enabling.
// - microsoft/Phi-4-multimodal-instruct -> microsoft/phi-4-multimodal-instruct
{
hfModel: "openai/whisper-large-v3",
providerModel: "openai/whisper:8099696689d249cf8b122d833c36ac3f75505c666a395ca40ef26f68e7d3d16e",
task: "automatic-speech-recognition",
},
];