diff --git a/app/kv-cache/KvCacheCalc.tsx b/app/kv-cache/KvCacheCalc.tsx index e80a22e..d92200d 100644 --- a/app/kv-cache/KvCacheCalc.tsx +++ b/app/kv-cache/KvCacheCalc.tsx @@ -111,7 +111,7 @@ export default function KvCacheCalc() { }; const catalogMatch = MODEL_OPTIONS.includes(model) - const kvModelStatus: ModelStatus = getAppConfig().supportedModels.includes(model) + const kvModelStatus: ModelStatus = getAppConfig().testedModels.includes(model) ? 'supported' : catalogMatch ? 'catalog' : catalogLoading ? 'fetching' diff --git a/app/performance/PerformanceEstimate.tsx b/app/performance/PerformanceEstimate.tsx index 14bf623..2eb9c70 100644 --- a/app/performance/PerformanceEstimate.tsx +++ b/app/performance/PerformanceEstimate.tsx @@ -39,8 +39,7 @@ import type { InferenceConfigResult } from '@/lib/gpu-math/inference-config'; import Link from 'next/link'; function modelSuggestions(): string { - const names = getAppConfig().suggestedModelNames; - return names.length > 0 ? names.join(', ') : 'Nemotron, DeepSeek V4, Gemma 4, Kimi'; + return getAppConfig().suggestedModelNames.join(', '); } function gpuOptionLabel(label: string, vramGb: number | null): string { @@ -114,7 +113,7 @@ export default function QuickEstimate() { const [isUsingFallback, setIsUsingFallback] = React.useState(false); const [fallbackReason, setFallbackReason] = React.useState(''); - const modelStatus: ModelStatus = getAppConfig().supportedModels.includes(model) + const modelStatus: ModelStatus = getAppConfig().testedModels.includes(model) ? 'supported' : aicModels.includes(model) ? 'catalog' @@ -227,7 +226,7 @@ export default function QuickEstimate() { setFallbackReason(''); // Skip HF fetch for supported and catalog models — we know they work - if (getAppConfig().supportedModels.includes(model) || aicModels.includes(model)) { + if (getAppConfig().testedModels.includes(model) || aicModels.includes(model)) { setIsFetchingConfig(false); return; } diff --git a/app/recommend/AdvancedEstimate.tsx b/app/recommend/AdvancedEstimate.tsx index 31c10fa..20f8d54 100644 --- a/app/recommend/AdvancedEstimate.tsx +++ b/app/recommend/AdvancedEstimate.tsx @@ -238,7 +238,7 @@ export default function AdvancedEstimate() { if (catalogLoading) { setModelStatus('idle'); return; } const timer = setTimeout(() => { if (!model.includes('/')) { setModelStatus('idle'); return; } - if (getAppConfig().supportedModels.includes(model)) { setModelStatus('supported'); return; } + if (getAppConfig().testedModels.includes(model)) { setModelStatus('supported'); return; } const inCatalog = MODEL_OPTIONS.includes(model); if (inCatalog) { setModelStatus('catalog'); return; } setModelStatus('fetching'); diff --git a/app/settings/Settings.tsx b/app/settings/Settings.tsx index 6e857b2..9cb1e97 100644 --- a/app/settings/Settings.tsx +++ b/app/settings/Settings.tsx @@ -55,7 +55,7 @@ export function Settings() { if (catalogLoading) { setModelStatus('idle'); return; } if (!localModel || !localModel.includes('/')) { setModelStatus('idle'); return; } const timer = setTimeout(() => { - if (getAppConfig().supportedModels.includes(localModel)) { setModelStatus('supported'); return; } + if (getAppConfig().testedModels.includes(localModel)) { setModelStatus('supported'); return; } const inCatalog = modelOptions.includes(localModel); if (inCatalog) { setModelStatus('catalog'); return; } setModelStatus('fetching'); @@ -143,12 +143,12 @@ export function Settings() { Models tested for use with the AIConfigurator sizing engine. - + {validatedOpen && (
- {getAppConfig().supportedModels.map(m => ( + {getAppConfig().testedModels.map(m => (
- {helperText ?? (supportedOnly ? ( + {helperText ?? (testedOnly ? ( Tested: {suggestedNames()}, ... — type to autocomplete ) : ( <>
Tested: {suggestedNames()}, ... — type to autocomplete
- {model && !supportedModels.includes(model) && cfg.modelRequestUrl && ( + {model && !testedModels.includes(model) && cfg.modelRequestUrl && (
New model? Request testing →
)} {hfToken ? ( diff --git a/lib/app-config.ts b/lib/app-config.ts index dd43d74..90b68ce 100644 --- a/lib/app-config.ts +++ b/lib/app-config.ts @@ -7,7 +7,7 @@ export interface AppConfig { defaultFrontierModel: string; defaultBackend: string; backendVersions: Record; - supportedModels: string[]; + testedModels: string[]; suggestedModelNames: string[]; modelRequestUrl: string; workloadPresets: WorkloadPreset[]; @@ -25,7 +25,7 @@ const FALLBACK: AppConfig = { 'tensorrt-llm': '11.2', 'sglang': '0.5.17', }, - supportedModels: [], + testedModels: [], suggestedModelNames: [], modelRequestUrl: '', workloadPresets: [], @@ -46,7 +46,7 @@ export async function loadAppConfig(): Promise { defaultFrontierModel: data.defaultFrontierModel ?? FALLBACK.defaultFrontierModel, defaultBackend: data.defaultBackend ?? FALLBACK.defaultBackend, backendVersions: data.backendVersions ?? FALLBACK.backendVersions, - supportedModels: data.supportedModels ?? FALLBACK.supportedModels, + testedModels: data.testedModels ?? FALLBACK.testedModels, suggestedModelNames: data.suggestedModelNames ?? FALLBACK.suggestedModelNames, modelRequestUrl: data.modelRequestUrl ?? FALLBACK.modelRequestUrl, workloadPresets: data.workloadPresets ?? FALLBACK.workloadPresets, diff --git a/public/config.json b/public/config.json index 565009e..326da6b 100644 --- a/public/config.json +++ b/public/config.json @@ -9,26 +9,17 @@ "tensorrt-llm": "11.2", "sglang": "0.5.17" }, - "suggestedModelNames": ["Gemma 4", "Nemotron Super", "DeepSeek V4", "Kimi K3"], - "supportedModels": [ - "google/gemma-4-26B-A4B", - "nvidia/Llama-3_3-Nemotron-Super-49B-v1", - "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4", - "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", - "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-FP8", - "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4", - "openai/gpt-oss-120b", - "openai/gpt-oss-20b", - "zai-org/GLM-5.2", - "zai-org/GLM-5.2-FP8", - "nvidia/GLM-5.2-NVFP4", - "deepseek-ai/DeepSeek-V4-Flash", - "sgl-project/DeepSeek-V4-Flash-FP8", + "suggestedModelNames": ["Gemma 4", "Nemotron Super", "DeepSeek V4", "gpt-oss"], + "testedModels": [ + "google/gemma-4-31B-it", + "RedHatAI/gemma-4-26B-A4B-it", + "RedHatAI/gemma-4-26B-A4B-it-FP8-dynamic", + "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8", + "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-FP8-block", "deepseek-ai/DeepSeek-V4-Pro", - "sgl-project/DeepSeek-V4-Pro-FP8", + "openai/gpt-oss-120b", + "RedHatAI/GLM-5.2-FP8", "moonshotai/Kimi-K2.5", - "nvidia/Kimi-K2.5-NVFP4", - "moonshotai/Kimi-K3" ], "modelRequestUrl": "https://github.com/redhat-performance/configiq/issues/new?labels=model-request&title=[Model+Request]+", "workloadPresets": [