Reserve Now Portal
Benchmarks

M5 Ultra Mac Studio: estimated LLM inference speeds

Tokens/sec estimates for eight open-weight models on the 256GB configuration, at the quantization level each model is commonly run at.

Updated September 2026
About these numbers

The figures below are third-party estimates, not measurements taken on maccloud's own hardware. They come from ModelFit's published bandwidth-and-model-size modeling for a 256GB M5 Ultra Mac Studio, whose own methodology note states plainly: "Speeds are ModelFit estimates from chip bandwidth and model size, not measured benchmarks." We're citing them because they're a reasonable, transparent starting point — not presenting them as in-house benchmark results. As we run our own measured benchmarks on maccloud hardware, we'll publish those separately and label them accordingly.

Estimated throughput by model

All figures below assume the 256GB configuration and the quantization level shown. Memory footprint includes ModelFit's KV-cache assumptions (FP16 cache, the llama.cpp/Ollama default) on top of model weights.

ModelParamsQuantizationMemory footprintEst. tokens/secEst. first token
Qwen3-Next 80B-A3B80BQ4_K_M~50.4 GB~64 tok/s~1.4s
GPT-OSS 120B117BMXFP4~65.4 GB~43 tok/s~1.5s
Qwen3-Next 80B-A3B80BQ8_0~84.8 GB~35 tok/s~1.5s
Qwen3.8-Flash-Next125BIQ1_S~123 GB~32 tok/s~1.6s
Laguna S 2.1118BQ4_K_M~96 GB~32 tok/s~1.6s
Qwen3.5 122B-A10B Instruct122BQ4_K_M~72 GB~28 tok/s~1.6s
Llama 4 Scout109BQ4_K_M~67 GB~23 tok/s~1.7s
Qwen3 235B A22B235BQ4_K_M~130 GB~14 tok/s~2.0s
Source: modelfit.io/mac-studio/m5, retrieved September 2026. Bandwidth-based estimates, not measured benchmarks — actual throughput varies by inference engine, context length, and prompt characteristics.

Where Llama 70B fits in

Llama 3.3 70B isn't in ModelFit's top-8 list for the 256GB config (it's tuned toward the larger MoE models that make better use of 256GB of headroom), but its own published GGUF sizes give a good proxy: at Q4_K_M (42.5GB) it sits in roughly the same memory class as GPT-OSS 120B above, and the same bandwidth-based reasoning suggests throughput in a broadly comparable range — likely faster than the 235B-class models on this list and slower than smaller 70–80B MoE models, given Llama 70B is a dense (non-MoE) architecture. See our Llama 70B sizing guide for the full quantization breakdown.

Why this matters for configuration choice

Every model on this list fits inside the 256GB plan with memory to spare. The 512GB plan exists for the cases that don't fit here: running the largest MoE models (like Qwen3 235B) at higher quantization for better quality, running two of these models concurrently, or leaving headroom for a long-context workload that pushes the KV cache well past what these estimates assume.

Reserve your M5 Ultra Mac Studio

256GB or 512GB unified memory, single-tenant, no billing until it ships.

Reserve Now