Can a Strix Halo Framework Desktop, 192GB run local LLMs?

Yes — 42 of the 48 current models on this site fit in its 160 GB of usable memory. The strongest it holds, Qwen3.8 Flash Next, runs at 31 tok/s with a 32k window, worked out from memory bandwidth. Whether that saves you money is a different question, and with no price announced all that can be said is the most it could cost and still come back: $130 inside two years at 500k tokens a day, on Ornith 1.5 122B, the model of the 42 it holds that saves the most.

Run the numbers on this machine

What the Framework Desktop, 192GB runs

They do not all hold the same window. The weights are a fixed size, but the key-value cache grows with every token you keep, so what is left of 160 GB after the weights is how far the context goes. On one of the 12 below, this machine's memory is what runs out first: MiniMax M2.7 stops soonest, at 64k. The rest reach 256k, where the calculator's list ends, or the longest setting their own context limit allows.

Speed and memory are at 32k context, the setting the calculator starts on; the memory column is the weights plus the cache for that much of it. Each speed says how it was arrived at, and every one here is estimated from this machine's 273 GB/s of memory bandwidth rather than taken from a published benchmark. There is a page on how that sum works, and what each size needs, and one on what 192 GB runs, machine by machine. The longest context is the longest setting the calculator offers that this machine still holds the model at, cache included, and each one opens the calculator on that model at that length. A figure tagged memory is one this machine ran out of room for, and the rest are stopped by the model's own limit or by the end of the list.

For scale, the calculator starts from 80 tok/s for a hosted API and times a local machine against it. Nothing above reaches it, and the quickest is 60 tok/s.

30 more fit. Ranked below the twelve above: Granite 4.2 30B, GLM-4.7-Flash, Gemma 4 12B, Qwen3.5 9B, Nemotron 3.5 Lightning 30B-A3B, MiniCPM5 2B, Qwen3.5 4B, gpt-oss-120b, Ling 3.0 tiny, Granite 4.2 8B, Mistral Small 4 (119B-2603), Qwen3-Coder 30B-A3B, Qwen3-Coder Next, gpt-oss-20b, Gemma 4 E4B, Devstral 2 123B, LFM2.5 2.6B, Devstral Small 2 24B, Ministral 3 14B, Ministral 3 8B. The other 10 have no intelligence-index score, so the ranking cannot place them: Laguna XS 2.1, LFM3 1.7B, Ornith 1.5 9B, Ornith 1.5 35B-A3B, Ornith 1.5 122B, Spark-X2.5 4B, Spark-X2.6 4B, KAT-Coder V2.5 Dev 35B-A3B, LFM2.5 24B-A2B, Flint 40B-A2.5B. Their pages show what each one needs and what runs it.

The 48 counted above are the current models. This site prices 16 superseded ones as well, because people still run them, and this machine holds 16 of those at 32k, strongest first: Qwen3 235B-A22B Instruct 2507, GLM-4.5-Air, Qwen3 30B-A3B Instruct 2507, DeepSeek-R1-Distill-Qwen-32B, DeepSeek-R1-Distill-Llama-70B, Llama 3.3 70B Instruct, Llama 3.1 8B Instruct at Q4_K_M, Llama 3.1 8B Instruct at Q8_0, Mistral Small 3.2 24B Instruct, Qwen3 32B at Q4_K_M, Qwen3 32B at Q8_0, Qwen3 14B, Llama 4 Scout 17B-16E, Qwen3 8B, Gemma 3 27B it, Gemma 3 12B it. Each of those pages asks whether the model is still worth running.

Other machines to weigh against the Framework Desktop, 192GB

Every row uses the same defaults as the figures above: 500k tokens a day at 15:1 input to output, 32k context, and each machine's strongest model that fits, counted against the same 48 current models.

The specifics

Chip

Framework Desktop — Ryzen AI Max+ PRO 495 'Gorgon Halo' · 192 GB

Memory bandwidth

273 GB/s — Measured GPU bandwidth is ~212-215 GB/s of the 256 GB/s theoretical, so estimates here run a little high.

Usable by the GPU

160 GB — 192 GB unified, of which AMD states 160 GB can be allocated to the GPU — the largest GPU-addressable pool of anything on this list under $10,000, if it ships at a sane price. Windows lets you dedicate up to 75% of RAM to the GPU (Variable Graphics Memory in AMD Adrenalin). On Linux the amdgpu GTT pool can be raised to roughly 108-120 GB with ttm.pages_limit / amdgpu.gttsize boot parameters. The Windows figure is used here as the conservative one.

Power under load

133 W (stand-in) — No measurement for Gorgon Halo; stand-in is the Framework Desktop's measured 133 W on the previous chip.

Availability

Announced, not yet priced or orderable.

Sources

Framework