The LPDDR5 memory is listed with a 16GB maximum; no RAM upgrade is assumed.
Listing evidence: 2026-09-25. Amazon.com base price visible in an owner-supplied screenshot reviewed 25 Sep 2026; capture time not supplied. Dated reference, not a live quote. Excludes card signup offers, coupons, tax and shipping.
Fixed 24GB memory and internal storage limit model size; no CUDA support; Amazon main-offer price and stock were not verified in this run.
Listing evidence: 2026-10-03. Apple US launch price announced 22 September 2026: USD 1,699. Dated manufacturer reference, not an Amazon offer; tax and shipping may differ.
Memory is soldered and shared with the operating system.
Listing evidence: 2026-09-24. GMKtec, EVO-X2 / 64GB / 1TB, recorded 22 Sep 2026; exact ASIN capacity identified 24 Sep. This is a dated reference, not a current Amazon price; tax, shipping and bundle contents may differ.
Weight size is not tested compatibility or speed. Check runtime support and memory for context before buying. Prices are dated references. As an Amazon Associate I earn from qualifying purchases.
How these options are selected
Up to three platforms, with the lowest dated reference price on each. Unknown prices follow known prices. Each memory pool keeps a reserve: 30% (minimum 2 GB) for system or unified memory; 20% (minimum 1 GB) for dedicated VRAM. System RAM is for CPU inference. This check does not establish GPU allocation, software support or context capacity.
Model file and details
Provider
Microsoft
Total parameters
4,741,793,792
Exact file size
2,716,078,176 bytes
Quantization / precision
Q4_K_M
Weight repository
Hansda/Mage-VL-Cowork-GGUF
Original model repository
microsoft/Mage-VL-Cowork
Weight revision
90c433e3d9ca1e6e0e8bf4bf3928a3a55a0b4a35
Model revision
6fbe7f91da477c1c73790dbb1f08d001b8626be1
Runtime notes
Confirm model architecture and quantization support in your runner.
Recorded release date
2026-10-30
Date evidence
HF source and third-party GGUF repositories created 30 October 2026; reviewed 6 November 2026. Creation date is an availability reference, not an independently verified launch date.
These scores were compiled from the linked sources. These published scores were not independently tested here. Different prompts, precision, thinking budgets and tools can change results.
OSWorld-Verified · Mage hybrid 361
66 / 100
Microsoft hybrid GUI and bash harness on 361 test_nogdrive tasks with a 50-step budget. Separate from Holo and other OSWorld protocols.
Benchmark efficiency: 24.3 published score points per GB of selected weights. This ratio is not an inference-speed measure; check the test settings to see whether the score covers the selected quantization.
Test settings: Microsoft source evaluation on 361 OSWorld-Verified test_nogdrive tasks, 50-step budget, hybrid computer_use and bash tools. SGLang /generate; thinking on, temperature 0.7, top_p 0.8, output cap 4,096 tokens; last three turns/screenshots retained.
Publisher-reported source-checkpoint result; not an independent test or a score measured on this Q4_K_M file. No confidence interval is supplied. Download bytes cover language weights, not total runtime memory. Apache-2.0. Custom Mage-ViT/llava_onevision2 vision architecture. Third-party Q4 decoder requires a separate projector for screenshots: Q8_0 projector adds 352,758,944 bytes (combined download 3,068,837,120 bytes); F16 alternative adds 661,433,504 bytes. Use the experimental Mage-enabled llama.cpp fork; stock llama.cpp, LM Studio and Ollama projector support is unvalidated. Use active 32,768-token context; 262,144 positions are not validated. Parameter count includes the source vision encoder. Browser-transfer 58.6 score belongs to SFT, not this RL checkpoint, and is excluded.
Context and version evidence: The repository and file name identify the selected model. Revision hashes are shown above when recorded. Tested context length, full hardware configuration and run date are not consistently recorded in this snapshot. Only settings explicitly stated in each benchmark record are known.
The 2.716 GB figure covers model weights in this file. It excludes the context cache, runtime, operating system and any separate multimodal files. It is not a measured RAM requirement or a token-generation speed.
Confirm model architecture and quantization support in your runner. The selected weights use 2.716 GB; this excludes context cache, runtime buffers and other programs.
These are the first listed tiers that pass our reserve policy, not measured minimum requirements. System RAM and VRAM remain separate. Use the memory tool to change the reserve for your context and runner.