# How much VRAM do you need for a 300B–480B model? > A 300B–480B LLM needs about 282.2 GB of VRAM at Q4, 493.9 GB at Q8 and 928.8 GB at FP16. Context-length table and the cheapest GPU and Mac that fit. Source: https://whatcanirunai.vercel.app/how-much-vram/400b/ # How much VRAM do you need for a 300B–480B model? Short answer: about **282.2 GB** at Q4\_K\_M, the usual quant for local use. Below: every quant, longer contexts, the cheapest hardware that fits, and every 300B–480B open model we track. Q4\_K\_M 282.2 GB 244.6 GB weights Q8\_0 493.9 GB 428.6 GB weights FP16 928.8 GB 806.8 GB weights Worked out for [Qwen3.5-397B-A17B](https://whatcanirunai.vercel.app/models/qwen3.5-397b-a17b/) (403.4B parameters, from the Hugging Face API). Formula: weights × 1.15 + 1 GB at a \~4k context ([how we calculate](https://whatcanirunai.vercel.app/method/)). No single consumer GPU we track fits it at Q4\. Cheapest Mac or mini PC: **Mac Studio M5 Ultra 512GB** ($22,299). ## Cheapest hardware for each quant Only devices with a sourced price. GPUs at list price (MSRP); street prices may be higher. __Cheapest GPU and Mac or mini PC for a 300B–480B model at each quant__ | Setting | Memory needed | Cheapest GPU | Cheapest Mac / mini PC | | ---------------------- | ------------- | ------------- | --------------------------------------------------------------------------------------- | | Q4\_K\_M, \~4k context | 282.2 GB | None we track | [Mac Studio M5 Ultra 512GB](https://whatcanirunai.vercel.app/hardware/mac-studio-m5-ultra-512/) 384 GB usable · $22,299 | | Q8\_0, \~4k context | 493.9 GB | None we track | None we track | | FP16, \~4k context | 928.8 GB | None we track | None we track | | Q4\_K\_M, 32k context | 283.1 GB | None we track | [Mac Studio M5 Ultra 512GB](https://whatcanirunai.vercel.app/hardware/mac-studio-m5-ultra-512/) 384 GB usable · $22,299 | No single consumer GPU we track holds a 300B–480B model at Q4\. Your options: two or more GPUs (see the [ready-made builds](https://whatcanirunai.vercel.app/builds/) or the [planner](https://whatcanirunai.vercel.app/planner/)), a large unified-memory machine (see the [best Mac for local LLMs](https://whatcanirunai.vercel.app/best-mac-for-local-llm/)), or a smaller quant with lower quality. ## Memory by context length Each extra token of context adds to the KV cache. Numbers for Qwen3.5-397B-A17B, from its published config; a Q8 KV cache halves the cache part. __Memory needed for a 300B–480B model by context length__ | Context | Q4\_K\_M, FP16 cache | Q4\_K\_M, Q8 cache | Q8\_0, FP16 cache | | ----------- | -------------------- | ------------------ | ----------------- | | 4k tokens | 282.2 GB | 282.2 GB | 493.9 GB | | 8k tokens | 282.4 GB | 282.2 GB | 494.0 GB | | 16k tokens | 282.6 GB | 282.4 GB | 494.3 GB | | 32k tokens | 283.1 GB | 282.6 GB | 494.8 GB | | 64k tokens | 284.1 GB | 283.1 GB | 495.8 GB | | 128k tokens | 286.1 GB | 284.1 GB | 497.8 GB | ## Hardware that runs it Bar: Q4 memory needed vs. usable memory. Speed is estimated decode at Q4 ([how we estimate](https://whatcanirunai.vercel.app/method/#speed)); MoE shows an upper bound only.FitsTight __Tracked hardware that runs a 300B–480B model at Q4__ | Hardware | Q4 need vs. usable | Q4 / Q8 / FP16 | Est. tok/s | | --------------------------------------------------------------------------------------- | -------------------- | ------------------------------ | ---------- | | [Mac Studio M5 Ultra 512GB](https://whatcanirunai.vercel.app/hardware/mac-studio-m5-ultra-512/)unified memory · $22,299 | **282.2** / 384.0 GB | Q4FitsQ8Won’t fitFP16Won’t fit | up to 93 | ## 300B–480B models we track __Memory needed by each 300B–480B model we track__ | Model | Q4\_K\_M | Q8\_0 | FP16 | | ----------------------------------------------------------------------------------------------------------------- | -------- | -------- | --------- | | [ERNIE-4.5-300B-A47B-PT](https://whatcanirunai.vercel.app/models/ernie-4.5-300b-a47b-pt/)300.5B params · baidu · MoE | 210.5 GB | 368.1 GB | 692.1 GB | | [DeepSeek-V4-Flash-0731](https://whatcanirunai.vercel.app/models/deepseek-v4-flash-0731/)304.2B params · deepseek-ai · MoE | 213.1 GB | 372.7 GB | 700.6 GB | | [DeepSeek-V4-Flash-Vision-Exp](https://whatcanirunai.vercel.app/models/deepseek-v4-flash-vision-exp/)304.6B params · deepseek-ai · MoE | 213.4 GB | 373.2 GB | 701.7 GB | | [MiMo-V2-Flash](https://whatcanirunai.vercel.app/models/mimo-v2-flash/)309.8B params · XiaomiMiMo · MoE | 217.0 GB | 379.5 GB | 713.5 GB | | [MiMo-V2.5](https://whatcanirunai.vercel.app/models/mimo-v2.5/)310.8B params · XiaomiMiMo · MoE | 217.7 GB | 380.7 GB | 715.7 GB | | [MiMo-V2.6-Flash-RL](https://whatcanirunai.vercel.app/models/mimo-v2.6-flash-rl/)310.8B params · XiaomiMiMo · MoE | 217.7 GB | 380.7 GB | 715.7 GB | | [MiMo-V2.6-Flash-MOPD](https://whatcanirunai.vercel.app/models/mimo-v2.6-flash-mopd/)310.8B params · XiaomiMiMo · MoE | 217.7 GB | 380.7 GB | 715.7 GB | | [Motif-3](https://whatcanirunai.vercel.app/models/motif-3/)314.8B params · Motif-Technologies | 220.5 GB | 385.7 GB | 725.1 GB | | [GLM-5.3-Flash](https://whatcanirunai.vercel.app/models/glm-5.3-flash/)321.3B params · zai-org · MoE | 225.0 GB | 393.6 GB | 740.0 GB | | [GLM-4.6](https://whatcanirunai.vercel.app/models/glm-4.6/)356.8B params · zai-org · MoE | 249.7 GB | 436.9 GB | 821.6 GB | | [GLM-4.5](https://whatcanirunai.vercel.app/models/glm-4.5/)358.3B params · zai-org · MoE | 250.8 GB | 438.8 GB | 825.2 GB | | [GLM-4.7](https://whatcanirunai.vercel.app/models/glm-4.7/)358.3B params · zai-org · MoE | 250.8 GB | 438.8 GB | 825.2 GB | | [Nex-N2.5-Pro](https://whatcanirunai.vercel.app/models/nex-n2.5-pro/)396.8B params · nex-agi · MoE | 277.6 GB | 485.8 GB | 913.6 GB | | [Nex-N2-Pro](https://whatcanirunai.vercel.app/models/nex-n2-pro/)396.8B params · nex-agi · MoE | 277.6 GB | 485.8 GB | 913.6 GB | | [NorthCUA-Pro](https://whatcanirunai.vercel.app/models/northcua-pro/)396.8B params · beipo-ai · MoE | 277.6 GB | 485.8 GB | 913.6 GB | | [Trinity-Large-Thinking](https://whatcanirunai.vercel.app/models/trinity-large-thinking/)398.6B params · arcee-ai · MoE | 278.9 GB | 488.1 GB | 917.9 GB | | [Trinity-Large-Preview](https://whatcanirunai.vercel.app/models/trinity-large-preview/)398.6B params · arcee-ai · MoE | 278.9 GB | 488.1 GB | 917.9 GB | | [Trinity-Large-TrueBase](https://whatcanirunai.vercel.app/models/trinity-large-truebase/)398.6B params · arcee-ai · MoE | 278.9 GB | 488.1 GB | 917.9 GB | | [Trinity-Large-Base](https://whatcanirunai.vercel.app/models/trinity-large-base/)398.6B params · arcee-ai · MoE | 278.9 GB | 488.1 GB | 917.9 GB | | [Llama-4-Maverick-17B-128E-Instruct](https://whatcanirunai.vercel.app/models/llama-4-maverick-17b-128e-instruct/)401.6B params · meta-llama · MoE | 281.0 GB | 491.7 GB | 924.6 GB | | [Qwen3.5-397B-A17B](https://whatcanirunai.vercel.app/models/qwen3.5-397b-a17b/)403.4B params · Qwen · MoE | 282.2 GB | 493.9 GB | 928.8 GB | | [Intern-S2-397B](https://whatcanirunai.vercel.app/models/intern-s2-397b/)403.4B params · internlm · MoE | 282.3 GB | 493.9 GB | 928.9 GB | | [ERNIE-4.5-VL-424B-A47B-PT](https://whatcanirunai.vercel.app/models/ernie-4.5-vl-424b-a47b-pt/)423.5B params · baidu · MoE | 296.3 GB | 518.5 GB | 975.1 GB | | [MiniMax-M3](https://whatcanirunai.vercel.app/models/minimax-m3/)427.0B params · MiniMaxAI · MoE | 298.7 GB | 522.8 GB | 983.2 GB | | [MiniMax-M1-40k](https://whatcanirunai.vercel.app/models/minimax-m1-40k/)456.1B params · MiniMaxAI · MoE | 319.0 GB | 558.3 GB | 1050.0 GB | | [MiniMax-M1-80k](https://whatcanirunai.vercel.app/models/minimax-m1-80k/)456.1B params · MiniMaxAI · MoE | 319.0 GB | 558.3 GB | 1050.0 GB | | [Qwen3-Coder-480B-A35B-Instruct](https://whatcanirunai.vercel.app/models/qwen3-coder-480b-a35b-instruct/)480.2B params · Qwen · MoE | 335.8 GB | 587.7 GB | 1105.4 GB | Have a different amount of memory? Try the [VRAM calculator](https://whatcanirunai.vercel.app/vram-calculator/) or see [best LLMs by VRAM size](https://whatcanirunai.vercel.app/best-llm-for/). ## Quick answers How much VRAM do I need for a 300B–480B model? About 282.2 GB at Q4\_K\_M, 493.9 GB at Q8\_0 and 928.8 GB at FP16, at a \~4k context. That is for Qwen3.5-397B-A17B (403.4B parameters): weights plus about 15% for the KV cache and 1 GB for the runtime. How much VRAM does a 300B–480B model need for 32k context? About 283.1 GB at Q4\_K\_M with an FP16 KV cache, or 282.6 GB with a Q8 cache, based on Qwen3.5-397B-A17B's published config. Models with different attention layouts need more or less. What is the cheapest GPU for a 300B–480B model? None of the single consumer GPUs we track hold it at Q4\. Use several GPUs, a large unified-memory machine, or CPU offload (slow). Can a Mac run a 300B–480B model? Yes. The cheapest Mac or unified-memory mini PC we track that runs it at Q4 is the Mac Studio M5 Ultra 512GB (512 GB, about 384.0 GB usable, $22,299). Does a MoE model of the same size need less VRAM? No. A mixture-of-experts model still has to hold all of its weights in memory, so it needs about the same VRAM as a dense model with the same total parameters. It runs faster, because only a few experts are read per token. Other sizes: [1B–2B](https://whatcanirunai.vercel.app/how-much-vram/1b/) · [2B–3B](https://whatcanirunai.vercel.app/how-much-vram/3b/) · [4B–6B](https://whatcanirunai.vercel.app/how-much-vram/4b/) · [7B–8B](https://whatcanirunai.vercel.app/how-much-vram/8b/) · [12B–14B](https://whatcanirunai.vercel.app/how-much-vram/14b/) · [16B–21B](https://whatcanirunai.vercel.app/how-much-vram/20b/) · [24B–27B](https://whatcanirunai.vercel.app/how-much-vram/27b/) · [30B–32B](https://whatcanirunai.vercel.app/how-much-vram/32b/) · [40B–50B](https://whatcanirunai.vercel.app/how-much-vram/49b/) · [70B](https://whatcanirunai.vercel.app/how-much-vram/70b/) · [120B](https://whatcanirunai.vercel.app/how-much-vram/120b/) · [200B–300B](https://whatcanirunai.vercel.app/how-much-vram/235b/) · [500B–800B](https://whatcanirunai.vercel.app/how-much-vram/671b/) · [1T+](https://whatcanirunai.vercel.app/how-much-vram/1t/) · [All sizes](https://whatcanirunai.vercel.app/how-much-vram/) · [Ollama VRAM by tag](https://whatcanirunai.vercel.app/ollama-vram-requirements/) · [GGUF sizes by quant](https://whatcanirunai.vercel.app/gguf-vram-requirements/) --- Agent guide: https://whatcanirunai.vercel.app/llms.txt · JSON API: https://whatcanirunai.vercel.app/developers/