AMD unified memory · Framework
Framework Desktop (Ryzen AI Max+ PRO 495)
Framework Desktop with AMD's Gorgon Halo chip and 192GB of memory, on pre-order with first batches shipping November 2026. The extra 64GB over Strix Halo machines is aimed at the large MoE models that do not fit in 128GB.
Memory for models
AMD documents 96 GB only for 128 GB machines; this applies AMD's "Custom (75%)" Variable Graphics Memory setting to 192 GB.
Memory bandwidth
Maker price
Largest at 4-bit
132.7 GB at 8,192 tokens of context (an assumed limit, see Memory for models)
gpt-oss-20b, 4-bit
no measurement on this hardware yet
Memory configurations
One page covers every memory option the maker sells. Prices are the maker’s list price for the cheapest configuration with that much memory.
| Memory | Price | Memory the GPU can use |
|---|---|---|
| 192 GB | ~$6,799 | 144 GB gpu-usable memoryassumption |
AMD documents 96 GB only for 128 GB machines; this applies AMD's "Custom (75%)" Variable Graphics Memory setting to 192 GB. How GPU memory limits work on unified-memory machines.
What it runs at 4-bit
Every model in the catalog at its 4-bit file with 8,192 tokens of context. 31 of 41 fit. Speeds are the median of measured runs on the same chip where any exist; otherwise an estimate from memory bandwidth. Try other quants and context lengths.
| Model | Needs | Fit | Generation speed |
|---|---|---|---|
| Kimi K32.8T, 104B active · UD-Q4_K_XL | 1406.8 GBcontext memory estimated | Does not fit | |
| DeepSeek V4 Pro1.6T, 49B active · Q4_K_M | 890.5 GBcontext memory not counted | Does not fit | |
| Kimi K2.61T, 32B active · UD-Q4_K_XL | 545.7 GBcontext memory estimated | Does not fit | |
| DeepSeek V4.1 Flash552B, 16B active · Q4 (DwarfStar) | 484.5 GB | Does not fit | |
| GLM-5.3753.3B · UD-Q4_K_XL | 437.4 GBcontext memory estimated | Does not fit | |
| DeepSeek V3.2671B, 37B active · Q4_K_M | 379.6 GBcontext memory estimated | Does not fit | |
| MiniMax M3428B, 23B active · Q4_K_M | 244.8 GBcontext memory not counted | Does not fit | |
| Qwen3.5 397B-A17B397B, 17B active · Q4_K_M | 229.1 GBcontext memory estimated | Does not fit | |
| GLM-5.3 Flash320B, 18B active · UD-Q4_K_XL | 187.6 GBcontext memory estimated | Does not fit | |
| Hy3295B, 21B active · Q4_K_M | 173.6 GB | Does not fit | |
| MiniMax M2.7228.7B · Q4_K_M | 132.7 GB | Fitsassumed limit | – |
| Qwen3.8 Flash-Next125B, 6B active · Q4_K_M | 113.1 GBcontext memory estimated | Fitsassumed limit | ~22–29 tok/sestimate |
| Nemotron 3 Super 120B-A12B120B, 12B active · Q4_K_M | 82.6 GBcontext memory estimated | Fitsassumed limit | ~15–19 tok/sestimate |
| Qwen3.5 122B-A10B122B, 10B active · Q4_K_M | 73.0 GBcontext memory estimated | Fitsassumed limit | ~20–26 tok/sestimate |
| Mistral Medium 3.5 128B128B · Q4_K_M | 74.0 GB | Fitsassumed limit | ~2.5–3.0 tok/sestimate |
| GLM-4.5-Air106B, 12B active · Q4_K_M | 70.9 GB | Fitsassumed limit | ~16–20 tok/sestimate |
| Mistral Small 4 119B (2603)119B, 6.5B active · Q4_K_M | 69.3 GBcontext memory estimated | Fitsassumed limit | ~32–41 tok/sestimate |
| Llama 4 Scout 17B-16E Instruct109B, 17B active · Q4_K_M | 62.4 GBcontext memory not counted | Fitsassumed limit | ~13–16 tok/sestimate |
| gpt-oss-120b117B, 5.1B active · MXFP4 (native) | 60.2 GBcontext memory estimated | Fitsassumed limit | ~47–60 tok/sestimate |
| Qwen3 Coder Next80B, 3B active · Q4_K_M | 46.9 GBcontext memory estimated | Fitsassumed limit | ~71–90 tok/sestimate |
| Llama 3.3 70B Instruct70B · Q4_K_M | 43.6 GB | Fitsassumed limit | ~4.4–5.2 tok/sestimate |
| Nemotron 3.5 Lightning 30B-A3B30B, 3B active · Q4_K_M | 25.3 GBcontext memory estimated | Fitsassumed limit | ~50–64 tok/sestimate |
| Qwen3.6 35B-A3B35B, 3B active · Q4_K_M | 22.4 GBcontext memory estimated | Fitsassumed limit | ~67–86 tok/sestimate |
| Ornith 1.5 35B-A3B35B, 3B active · Q4_K_M | 21.9 GBcontext memory estimated | Fitsassumed limit | ~69–88 tok/sestimate |
| Qwen3 Coder 30B-A3B Instruct30.5B, 3.3B active · Q4_K_M | 19.5 GB | Fitsassumed limit | ~64–82 tok/sestimate |
| Gemma 4 31B IT30.7B · Q4_K_M | 18.9 GBcontext memory estimated | Fitsassumed limit | ~10–12 tok/sestimate |
| Granite 4.2 30B29.3B · Q4_K_M | 20.3 GB | Fitsassumed limit | ~10–12 tok/sestimate |
| Qwen3.8 27B27B · Q4_K_M | 18.2 GBcontext memory estimated | Fitsassumed limit | ~11–13 tok/sestimate |
| Gemma 4 26B-A4B IT25.2B, 3.8B active · Q4_K_M | 17.4 GBcontext memory estimated | Fitsassumed limit | ~50–64 tok/sestimate |
| Qwen3.6 27B27B · Q4_K_M | 17.7 GBcontext memory estimated | Fitsassumed limit | ~11–13 tok/sestimate |
| Devstral Small 2 24B (2512)24B · Q4_K_M | 16.1 GB | Fitsassumed limit | ~13–15 tok/sestimate |
| gpt-oss-20b21B, 3.6B active · MXFP4 (native) | 12.5 GBcontext memory estimated | Fitsassumed limit | ~65–82 tok/sestimate |
| Phi-414.7B · Q4_K_M | 11.3 GB | Fitsassumed limit | ~21–25 tok/sestimate |
| Ministral 3 14B Reasoning (2512)13.9B · Q4_K_M | 10.4 GB | Fitsassumed limit | ~22–27 tok/sestimate |
| Gemma 4 12B IT11.95B · Q4_K_M | 8.2 GBcontext memory estimated | Fitsassumed limit | ~26–31 tok/sestimate |
| Qwen3.5 9B9B · Q4_K_M | 7.0 GBcontext memory estimated | Fitsassumed limit | ~33–39 tok/sestimate |
| Granite 4.2 8B8.8B · Q4_K_M | 7.9 GB | Fitsassumed limit | ~33–40 tok/sestimate |
| Gemma 4 E4B IT8B · Q4_K_M | 6.5 GBcontext memory not counted | Fitsassumed limit | ~61–73 tok/sestimate |
| Qwen3 8B8.2B · Q4_K_M | 7.3 GB | Fitsassumed limit | ~37–44 tok/sestimate |
| Llama 3.1 8B Instruct8B · Q4_K_M | 7.1 GB | Fitsassumed limit | ~38–45 tok/sestimate |
| Qwen3.5 4B4B · Q4_K_M | 4.3 GBcontext memory estimated | Fitsassumed limit | ~68–81 tok/sestimate |
Measured speeds
Nobody has published a run on this chip with a stated model, quant and backend that we could find. The speeds above are estimates.
Specifications
- Chip
- Ryzen AI Max+ PRO 495
- Memory
- LPDDR5X-8533, 256-bit, soldered
- Memory bandwidth
- 273 GB/s
- FP16 compute
- AMD quotes up to 132 overall TOPS (precision not stated) and no FP16 TFLOPS figure
- NPU
- 55 TOPS
- Software
- ROCm, Vulkan, DirectML
- Ships with
- None on DIY Edition; Fedora 44 KDE is the default on the pre-built per Phoronix
- Power
- 400 W (Internal 400W power supply (Framework spec))
- Form factor
- Mini PC
- Networking
- 5GbE (Realtek RTL8126); two USB4 ports; optional 10GbE expansion card
- Released
- 2026-11
- Maker page
- frame.work
Sources
- frame.work: $6,799, LPDDR5x-8533, batches ship Nov/Dec
- phoronix.com: pre-orders opened 2026-09-30, first shipments November, pre-built $7,449, Fedora default
- amd.com: Radeon 8065S 40 CU 3.0GHz, max 192GB LPDDR5x-8533, NPU 55 TOPS, 132 overall TOPS
- rocm.docs.amd.com: Linux GTT default ~50% of RAM; raise via ttm pages_limit / amd-ttm --set (AMD example: 100GB on a 128GB system); keep BIOS VRAM small