Low-power CPU and shared graphics limit heavy inference.
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 36GB memory and internal storage; no CUDA support; price is a manufacturer reference rather than a verified Amazon offer.
Listing evidence: 2026-10-02. Apple US launch price for the base M5 Max Mac Studio, published 22 September 2026: USD 2,499. 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
Nanbeige
Total parameters
33,118,977,024
Exact file size
20,788,825,216 bytes
Quantization / precision
Q4_K_M
Weight repository
bartowski/Nanbeige_Nanbeige4.3-33B-GGUF
Original model repository
Nanbeige/Nanbeige4.3-33B
Weight revision
9beb68421fdf7be095093a6888325fa2c83ddd38
Model revision
c1093c501cc6460427c0ac634db7223d0450c0ee
Runtime notes
Confirm model architecture and quantization support in your runner.
Recorded release date
2026-10-24
Date evidence
Source repository created 2026-10-24; availability basis, not a proven announcement date. Apache-2.0. Bartowski complete Q4_K_M language weights. Source scores are not measured quantized quality. Check runtime support; no vision projector or speculative draft is included.
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.
GPQA Diamond
89.5 / 100
Hard science questions. This is a science reasoning test, not an overall intelligence score.
Benchmark efficiency: 4.305 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: Highest thinking effort with preserve_thinking=true; reasoning settings temperature 0.6, top-p 0.95, max new tokens 262144. Repeat count not stated.
Publisher result; no independent replication or confidence interval supplied. Apache-2.0. Bartowski complete Q4_K_M language weights. Source scores are not measured quantized quality. Check runtime support; no vision projector or speculative draft is included. Weight size excludes runtime and KV-cache memory; this is not a measured device-fit or speed result.
Competition math from AIME 2026. Published answer accuracy.
Benchmark efficiency: 4.579 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: Highest thinking effort, preserve_thinking=true. Reasoning settings temperature 0.6, top-p 0.95, max new tokens 262144. Repeat count not stated.
Publisher result; no independent replication or confidence interval supplied. Apache-2.0. Bartowski complete Q4_K_M language weights. Source scores are not measured quantized quality. Check runtime support; no vision projector or speculative draft is included. Weight size excludes runtime and KV-cache memory; this is not a measured device-fit or speed result.
Agentic terminal-use task success. Publisher-reported percentages are retained, but agent scaffold, tools, timeouts and reasoning budget can materially affect results.
Benchmark efficiency: 3.892 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: Terminus 2, highest thinking effort, preserve_thinking=true; agent sampling temperature 1.0, top-p 0.95, max new tokens 65536. Repeat count not stated.
Publisher result; no independent replication or confidence interval supplied. Apache-2.0. Bartowski complete Q4_K_M language weights. Source scores are not measured quantized quality. Check runtime support; no vision projector or speculative draft is included. Weight size excludes runtime and KV-cache memory; this is not a measured device-fit or speed result.
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 20.789 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 20.789 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.