@@ -21,6 +21,7 @@ export const removeModelSuffix = (id: string) => {
2121 . replace ( / _ l l a m a f i l e _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } / , "" )
2222 . replace ( / _ o l l a m a 2 _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } / , "" )
2323 . replace ( / _ l l a m a c p p _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } / , "" )
24+ . replace ( / _ v l l m _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } / , "" )
2425}
2526export const isLMStudioModel = ( model : string ) => {
2627 const lmstudioModelRegex =
@@ -39,6 +40,11 @@ export const isLLamaCppModel = (model: string) => {
3940 return llamaCppModelRegex . test ( model )
4041}
4142
43+ export const isVLLMModel = ( model : string ) => {
44+ const vllmModelRegex = / _ v l l m _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } /
45+ return vllmModelRegex . test ( model )
46+ }
47+
4248export const isOllamaModel = ( model : string ) => {
4349 const ollamaModelRegex =
4450 / _ o l l a m a 2 _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } /
@@ -98,6 +104,20 @@ export const getLLamaCppModelId = (
98104 return null
99105}
100106
107+ export const getVLLMModelId = (
108+ model : string
109+ ) : { model_id : string ; provider_id : string } => {
110+ const vllmModelRegex =
111+ / _ v l l m _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } /
112+ const match = model . match ( vllmModelRegex )
113+ if ( match ) {
114+ const modelId = match [ 0 ]
115+ const providerId = match [ 0 ] . replace ( "_vllm_openai-" , "" )
116+ return { model_id : modelId , provider_id : providerId }
117+ }
118+ return null
119+ }
120+
101121export const isCustomModel = ( model : string ) => {
102122 if ( isLMStudioModel ( model ) ) {
103123 return true
@@ -115,6 +135,10 @@ export const isCustomModel = (model: string) => {
115135 return true
116136 }
117137
138+ if ( isVLLMModel ( model ) ) {
139+ return true
140+ }
141+
118142 const customModelRegex =
119143 / _ m o d e l - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 , 4 } - [ a - f 0 - 9 ] { 4 } /
120144 return customModelRegex . test ( model )
@@ -258,7 +282,7 @@ export const getModelInfo = async (id: string) => {
258282 if ( isLLamaCppModel ( id ) ) {
259283 const llamaCppId = getLLamaCppModelId ( id )
260284 if ( ! llamaCppId ) {
261- throw new Error ( "Invalid LMStudio model ID" )
285+ throw new Error ( "Invalid llamaCPP model ID" )
262286 }
263287
264288 return {
@@ -274,6 +298,24 @@ export const getModelInfo = async (id: string) => {
274298 }
275299 }
276300
301+ if ( isVLLMModel ( id ) ) {
302+ const vllmId = getVLLMModelId ( id )
303+ if ( ! vllmId ) {
304+ throw new Error ( "Invalid Vllm model ID" )
305+ }
306+ return {
307+ model_id : id . replace (
308+ / _ v l l m _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } / ,
309+ ""
310+ ) ,
311+ provider_id : `openai-${ vllmId . provider_id } ,
312+ name : id . replace (
313+ / _ v l l m _ o p e n a i - [ a - f 0 - 9 ] { 4 } - [ a - f 0 - 9 ] { 3 } - [ a - f 0 - 9 ] { 4 } / ,
314+ ""
315+ )
316+ }
317+ }
318+
277319
278320 if ( isOllamaModel ( id ) ) {
279321 const ollamaId = getOllamaModelId ( id )
@@ -388,6 +430,30 @@ export const dynamicFetchLLamaCpp = async ({
388430 return llamaCppModels
389431}
390432
433+
434+ export const dynamicFetchVLLM = async ( {
435+ baseUrl,
436+ providerId,
437+ customHeaders = [ ]
438+ } : {
439+ baseUrl : string
440+ providerId : string
441+ customHeaders ?: { key : string ; value : string } [ ]
442+ } ) => {
443+ const models = await getAllOpenAIModels ( { baseUrl, customHeaders } )
444+ const vllmModels = models . map ( ( e ) => {
445+ return {
446+ name : e ?. name || e ?. id ,
447+ id : `${ e ?. id } ${ providerId } ,
448+ provider : providerId ,
449+ lookup : `${ e ?. id } ${ providerId } ,
450+ provider_id : providerId
451+ }
452+ } )
453+
454+ return vllmModels
455+ }
456+
391457export const dynamicFetchOllama2 = async ( {
392458 baseUrl,
393459 providerId,
@@ -459,6 +525,10 @@ export const ollamaFormatAllCustomModels = async (
459525 ( model ) => model . provider === "llamacpp"
460526 )
461527
528+ const vllmProviders = allProviders . filter (
529+ ( model ) => model . provider === "vllm"
530+ )
531+
462532 const lmModelsPromises = lmstudioProviders . map ( ( provider ) =>
463533 dynamicFetchLMStudio ( {
464534 baseUrl : provider . baseUrl ,
@@ -489,6 +559,13 @@ export const ollamaFormatAllCustomModels = async (
489559 customHeaders : provider . headers
490560 } ) )
491561
562+ const vllmModelsPromises = vllmProviders . map ( ( provider ) =>
563+ dynamicFetchVLLM ( {
564+ baseUrl : provider . baseUrl ,
565+ providerId : provider . id ,
566+ customHeaders : provider . headers
567+ } ) )
568+
492569 const lmModelsFetch = await Promise . all ( lmModelsPromises )
493570
494571 const llamafileModelsFetch = await Promise . all ( llamafileModelsPromises )
@@ -497,6 +574,8 @@ export const ollamaFormatAllCustomModels = async (
497574
498575 const llamacppModelsFetch = await Promise . all ( llamacppModelsPromises )
499576
577+ const vllmModelsFetch = await Promise . all ( vllmModelsPromises )
578+
500579 const lmModels = lmModelsFetch . flat ( )
501580
502581 const llamafileModels = llamafileModelsFetch . flat ( )
@@ -505,6 +584,8 @@ export const ollamaFormatAllCustomModels = async (
505584
506585 const llamacppModels = llamacppModelsFetch . flat ( )
507586
587+ const vllmModels = vllmModelsFetch . flat ( )
588+
508589 // merge allModels and lmModels
509590 const allModlesWithLMStudio = [
510591 ...( modelType !== "all"
@@ -513,7 +594,8 @@ export const ollamaFormatAllCustomModels = async (
513594 ...lmModels ,
514595 ...llamafileModels ,
515596 ...ollama2Models ,
516- ...llamacppModels
597+ ...llamacppModels ,
598+ ...vllmModels
517599 ]
518600
519601 const ollamaModels = allModlesWithLMStudio . map ( ( model ) => {
0 commit comments