AMD
Operating mode
Use this to bias workload recommendations toward responsiveness, background autonomy, lighter serving, or multi-GPU scale-out.
Current mode
Balanced
Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.
The AMD Instinct MI300A 128GB is a unique APU-style CDNA 3 accelerator combining CPU cores (Zen 4) and GPU compute on the same package with a unified 128 GB HBM3 memory pool. Unlike the discrete MI300X, the MI300A's memory is shared between CPU and GPU — eliminating PCIe transfer overhead for AI workloads where CPU preprocessing and GPU inference must cooperate. It delivers 1.2 PFLOPS FP16 with 5.3 TB/s of memory bandwidth.
Beyond LLMs
What AI tasks this GPU can handle — from text generation to image and video creation.
| Capability | Status | Representative Model |
|---|---|---|
| LLM Chat (7B) | Runs natively | Llama 3.1 8B Q4 |
| LLM Coding (30B) | Runs natively | Qwen 3 30B Q4 |
| LLM Large (70B) | Runs natively | Llama 3.1 70B Q4 |
| Image Gen (SDXL) | Runs natively | SDXL 1.0 FP16 |
| Image Gen (Flux) | Runs natively | Flux.1 Dev FP16 |
| Image Gen (SD 3.5) | Runs natively | SD 3.5 Large FP16 |
| Video Short (25f) | Runs natively | LTX Video 2B |
| Video Long (100f) | Runs natively | Wan Video 14B |
Architecture
CDNA 3 powers the Instinct MI300X (GPU-only, 192 GB HBM3) and MI300A (APU with integrated CPU). It features advanced packaging with up to 12 chiplets and native FP8 support for AI inference.
AI Relevance
The MI300X with 192 GB HBM3 can hold even the largest open-weight models (70B+ at full precision) entirely in GPU memory. FP8 support and mature ROCm stack make it a serious competitor to NVIDIA H100 for AI inference.
Buying advice
Excellent choice for local AI
Runs 35 of 50 top models well — a strong all-rounder for local inference.
128.0 GB
VRAM
$12,000
MSRP
$94/GB
Cost per GB VRAM
Best models for this GPU
What will limit you first
This setup is broadly balanced for this model.
No major red flags
This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.
Best upgrade itinerary
Unlocks 2 additional models that do not fit on the current setup.
Want more headroom? NVIDIA H200 141GB (141.0 GB VRAM) is the next step up.
Chat
SQwen 3.5 122B A10B matches Chat and keeps a practical fit profile. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, lm-studio.
Coding
SQwen3-Coder-Next is a specialized fit for Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Agentic Coding
SDevstral 2 123B Instruct is a specialized fit for Agentic Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, lm-studio.
Reasoning
SDevstral 2 123B Instruct matches Reasoning and keeps a practical fit profile. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, lm-studio.
RAG
SQwen 3.5 122B A10B matches RAG and keeps a practical fit profile. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, lm-studio.
Qwen 3.5 122B A10B
Devstral 2 123B InstructMistral Small 4 119B
GPT-OSS 120B
Command A 111BPixtral Large 124BLeanstral 119B A6BQwen 2.5 VL 72BQwen3-Coder-NextQwen 3.6 35B A3BQwen3-Coder 30B A3B InstructQwen 3.5 27BQwen3-VL 30B A3B InstructQwen 3.5 35B A3BQwen 3.6 27BQwen 3 32BMagistral Small 2507Devstral Small 2 24B InstructQwen 3 30B A3B
Nemotron 3 Nano 30BQwen 3.5 9BQwen 3 14B
Gemma 4 31BDevstral Small 1.1
Phi-4-reasoning-plus 14BNemotron Cascade 2 30B A3BQwen 3 8BGPT-OSS 20BQwen 3.5 4B
EXAONE 4.0 32BMinistral 3 14BNemotron Nano 8BPhi-4 Mini Reasoning 4B
Jina Embeddings v3
BGE M3Qwen 3.5 397B A17B
Kimi K2.5Kimi K2.6
DeepSeek V4 ProDeepSeek V4 Flash
GLM-5.2GLM-5.1GLM-5DeepSeek V3.2Qwen 3 235B A22BQwen3-Coder 480B A35B InstructDeepSeek Coder V2 236BDeepSeek R1 671BDeepSeek V3.1 671BJust out of reach
High-quality models that need a bit more memory
Image & Video Generation
56 of 56 models can generate images or video on your AMD Instinct MI300A 128GB
| Model | Max Resolution | Gen Time | Grade |
|---|---|---|---|
| SD TurboImage | 512×512 | 0ms | S |
| Stable Diffusion 1.5Image | 512×768 | 100ms | S |
| Realistic Vision v5.1Image | 512×768 | 100ms | S |
| DreamShaper 8Image | 512×768 | 100ms | S |
| LCM DreamShaper v7Image | 512×768 | 0ms | S |
| PixArt-SigmaImage | 1024×1024 | 300ms | S |
| FramePack I2VVideo | 1280×720 | 500ms/frame | S |
| SDXL TurboImage | 512×512 | 0ms | S |
| SDXL LightningImage | 1024×1024 | 100ms | S |
| Stable Diffusion XL 1.0Image | 1024×1024 | 300ms | S |
| Playground v2.5Image | 1024×1024 | 400ms | S |
| RealVisXL v5.0Image | 1024×1024 | 300ms | S |
| DreamShaper XLImage | 1024×1024 | 300ms | S |
| Juggernaut XL v9Image | 1024×1024 | 300ms | S |
| Animagine XL 3.1Image | 1024×1024 | 300ms | S |
| Pony Diffusion V6 XLImage | 1024×1024 | 300ms | S |
| Animagine XL 4.0Image | 1024×1024 | 300ms | S |
| Illustrious XLImage | 1024×1024 | 300ms | S |
| Wan Video 2.1 1.3BVideo | 480×832 | 200ms/frame | S |
| Stable Diffusion 3.5 MediumImage | 1024×1024 | 500ms | S |
| Flux.2 Klein 4BImage | 1024×1024 | 100ms | S |
| LTX Video 2BVideo | 1280×720 | 300ms/frame | S |
| KolorsImage | 1024×1024 | 600ms | S |
| Stable CascadeImage | 1024×1024 | 700ms | S |
| AuraFlow v0.3Image | 1536×1536 | ~1.3s | S |
| Stable Diffusion 3.5 LargeImage | 1024×1024 | ~1.6s | S |
| Stable Diffusion 3.5 Large TurboImage | 1024×1024 | 300ms | S |
| CogVideoX 2BVideo | 720×480 | 300ms/frame | S |
| HunyuanVideoVideo | 720×1280 | 500ms/frame | S |
| ChromaImage | 1024×1024 | 300ms | S |
| Z-Image TurboImage | 1536×1536 | 300ms | S |
| Flux.1 DevImage | 1024×1024 | ~1.3s | S |
| Flux.1 SchnellImage | 1024×1024 | 300ms | S |
| LTX Video 13BVideo | 1280×720 | 500ms/frame | S |
| Flux.1 Kontext DevImage | 1024×1024 | ~1.5s | S |
| AnimateDiff v1.5.3Video | 512×768 | 100ms/frame | S |
| Cosmos Diffusion 7BVideo | 1024×576 | 400ms/frame | S |
| CogVideoX 5BVideo | 720×480 | 400ms/frame | S |
| Wan2.2 TI2V 5BVideo | 832×480 | 400ms/frame | S |
| Flux.2 Klein 9BImage | 1024×1024 | 100ms | S |
| Flux.1 Fill DevImage | 1024×1024 | ~1.3s | S |
| Krea 2Image | 1024×1024 | 400ms | S |
| Sulphur 2Video | 1280×720 | 500ms/frame | S |
| Ideogram 4Image | 2048×2048 | 400ms | S |
| Mochi 1 PreviewVideo | 848×480 | 500ms/frame | S |
| HunyuanVideo 1.5Video | 720×1280 | 500ms/frame | S |
| Helios 14BVideo | 1280×720 | 600ms/frame | S |
| SkyReels V2 14BVideo | 1280×720 | 600ms/frame | S |
| Wan Video 2.1 14BVideo | 720×1280 | 600ms/frame | S |
| Wan Video 2.2 14BVideo | 720×1280 | 600ms/frame | S |
| Qwen ImageImage | 1024×1024 | 500ms | S |
| Qwen Image EditImage | 1024×1024 | 500ms | S |
| LTX-2 22BVideo | 1920×1088 | 700ms/frame | S |
| Flux.2 DevImage | 1024×1024 | ~14s | S |
| MAGI-1Video | 1280×720 | 700ms/frame | S |
| HunyuanImage 3.0Image | 256×256 | 900ms | D |
Image models estimated at 1024×1024 (28 steps, FP16). Video models estimated at 768×512 (25 frames, 30 steps, FP16). Actual performance varies with runtime and system load.
Multi-GPU scaling
Scale out with multiple GPUs for larger models. Infinity Fabric provides 896 GB/s inter-GPU bandwidth with 12% overhead.
| Config | Effective memory | Models that fit | Est. bandwidth |
|---|---|---|---|
| 1× AMD | 128 GB | 356/380 | 5,300 GB/s |
| 2× AMD | 256 GB | 368/380 | 9,328 GB/s |
| 4× AMD | 512 GB | 377/380 | 18,656 GB/s |
Model counts use default quantization at coding workload settings. Multi-GPU scaling factor: 0.88× per additional GPU.
Upgrade paths
See what you unlock with more powerful hardware
Upgrade options
Unlocks 21 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 72%.
Infinity Fabric gives this scale-out path a cleaner inter-GPU story than PCIe-only builds.
~$12,000 MSRP
NVIDIA H200 141GBNext step upUnlocks 2 additional models that do not fit on the current setup.
~$30,000 MSRP
NVIDIA B200 180GBBiggest leapUnlocks 8 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 29%.
~$30,000 MSRP
Unlocks 12 additional models that do not fit on the current setup.
~$20,000 MSRP
Unlocks 13 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 20%.
~$8,000 MSRP
AMD Instinct MI300A 128GB (128 GB VRAM) can run these top models: Qwen 3.5 122B A10B (score: 99/100), Devstral 2 123B Instruct (score: 98/100), Mistral Small 4 119B (score: 97/100). See the full compatibility list above.
AMD Instinct MI300A 128GB has 128 GB of VRAM available for AI model inference. This determines which models and quantization levels you can run locally.
Yes, AMD Instinct MI300A 128GB is excellent for running LLMs locally with top compatibility scores above 80/100.
For coding on AMD Instinct MI300A 128GB, we recommend Qwen3-Coder-Next. It achieves 157.1 tokens per second with 256K context window. Qwen3-Coder-Next is a specialized fit for Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
There are 5 upgrade path(s) from AMD Instinct MI300A 128GB: AMD Instinct MI300A 128GB, NVIDIA H200 141GB. Upgrading would unlock larger models and faster inference speeds.
Yes, AMD Instinct MI300A 128GB with 128 GB of usable memory can run Flux.1 Dev at FP16 natively. Flux is a 12B parameter diffusion transformer that produces high-quality images. You can also run the Schnell variant for faster generation.
AMD Instinct MI300A 128GB (128 GB VRAM) can handle various AI generation tasks beyond LLMs. For image generation, SDXL and Stable Diffusion 3.5 run well. Flux.1 Dev also runs natively for state-of-the-art image quality. For video, LTX Video 2.3 can generate short clips. Check the AI Capability Matrix above for detailed compatibility.
AMD Instinct MI300A 128GB is excellent for AI image generation. With 128 GB of usable memory, it runs all major diffusion models including Flux.1, SDXL, and Stable Diffusion 3.5 at full precision. You can generate high-resolution images quickly and even handle video generation models.
Yes, AMD Instinct MI300A 128GB with 128 GB of usable memory can run Qwen 3.5 27B at Q8 (near-lossless, ~28.9 GB) or even FP16 (~55.4 GB) depending on your context needs. This setup provides an excellent experience with this model. Use Ollama or vLLM for best results.
With 128 GB VRAM on AMD Instinct MI300A 128GB, use Q8_0 for most models — it is near-lossless and you have the memory for it. For 70B+ models, Q6_K offers excellent quality. Reserve Q4_K_M for 100B+ models or when you need maximum context length.
AMD Instinct MI300A 128GB already has strong memory bandwidth, so the next limit is often memory capacity and context headroom rather than raw decode speed. For local LLMs, fit first and bandwidth second is the right mental model.
AMD Instinct MI300A 128GB supports up to 4× GPU scaling via Infinity Fabric at 896 GB/s. With 4× GPUs, you get 512 GB effective memory with a 0.88× scaling factor per GPU. This enables running models like Qwen 3.5 397B A17B and Kimi K2.5 that don't fit on a single card.
Infinity Fabric is recommended for AMD Instinct MI300A 128GB multi-GPU inference, providing 896 GB/s interconnect bandwidth with only 12% scaling overhead. PCIe-only setups work but have higher overhead (~25%) due to limited inter-GPU bandwidth.
Compare with similar
