← Files AMDARCHIVED FILE
skills/serving-llms-on-instinct/data/blacklist.json
2 KB · Oct 4, 2026 · 12:28 UTC
{
"_comment": "Models in vLLM recipes that cannot be served as LLM endpoints on AMD Instinct. The agent should refuse these and explain why.",
"not_an_llm": {
"_comment": "Non-LLM models: diffusion, image gen, audio gen, embeddings, rerankers. These are not chat/completion endpoints.",
"models": [
"stabilityai/stable-diffusion-3.5-medium",
"stabilityai/stable-audio-open-1.0",
"Wan-AI/Wan2.2-T2V-A14B-Diffusers",
"Wan-AI/Wan2.2-T2V-A14B",
"Qwen/Qwen-Image",
"zai-org/GLM-Image",
"zai-org/Glyph",
"meituan-longcat/LongCat-Image-Edit",
"jinaai/jina-embeddings-v5-text-small",
"jinaai/jina-reranker-m0",
"black-forest-labs/FLUX.1-dev",
"black-forest-labs/FLUX.2-dev",
"black-forest-labs/FLUX.2-klein-9B",
"Tongyi-MAI/Z-Image-Turbo",
"openai/whisper-large-v3-turbo",
"neuphonic/neutts-air"
]
},
"needs_audio_pipeline": {
"_comment": "Models that require audio input/output streaming. vLLM can load them but they need a specialized audio pipeline, not a standard chat endpoint.",
"models": [
"Qwen/Qwen3-ASR-1.7B",
"zai-org/GLM-ASR-Nano-2512",
"mistralai/Voxtral-Mini-4B-Realtime-2602",
"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16"
]
},
"needs_vllm_nightly": {
"_comment": "Models that require unreleased vLLM nightly. Will break on any stable Docker image. Re-check when vLLM cuts a new release.",
"models": [
"internlm/Intern-S2-Preview",
"mistralai/Mistral-Medium-3.5-128B",
"poolside/Laguna-XS.2",
"stepfun-ai/Step-3.7-Flash"
]
},
"translation_only": {
"_comment": "Translation-only models with no chat or tool-calling support.",
"models": [
"Google/translategemma-27b-it"
]
},
"incompatible_weights": {
"_comment": "Models whose weight naming or architecture is incompatible with current vLLM stable (v0.22.0). Crashes during weight loading.",
"models": [
"thu-pacman/PCMind-2.1-Kaiyuan-2B"
]
}
}
SHA-256: 59feb6605a72250100884dadd4d89473eda4deb3c2827fa6656a39276ef5d347