Road warrior with compromises - Lenovo ThinkPad X13 Gen 7 AMD Laptop Review

[Redaktion]

Lenovo's compact ThinkPad X13 Gen 7 keeps the lightweight chassis and gets new AMD processors, but the overall package is very familiar. The price, however, is much higher this year.

https://www.notebookcheck.net/Road-warrior-with-compromises-Lenovo-ThinkPad-X13-Gen-7-AMD-Laptop-Review.1395047.0.html

[xyzax]

Why are you insisting on shaving AMD s devices Cinebench R15 single thread score by 25% ?..
It seems that it s an habit as it often happen only with AMD and never with an Intel device,
the abnormal score should be noticed since all other single thread tests are accurate,
wich point to a flawed methodology, the stated score here amount to no more than 3.6GHz
in the CB R15 single thread score.

[32GB RAM not for local AI]

RAM not upgradable (maximum 32 GB on AMD SKUs)

Wow, I clicked to look up something else, not a 32 GB RAM limit! Well, then, "here we go again":
With only 32 GB of total memory you can fit about this many context tokens of SOTA (in its class size) AI LLM model[1] huggingface.co/unsloth/Qwen3.6-35B-A3B-MTP-GGUF 4-bit UD-Q4_K_XL quant:

.. I've come to the conclusion that (1) 32768 is the biggest context I can get away with in an adequately smart model, and (2) it just ain't enough.

(1 token = 0.75 words)

It's also possible to calculate the available context[2]: 16,384 [KV cache context tokens per GB]*(32 [GB total memory] - 6 to 8 [GB for the OS] - [GB quant filesize]):
~34,400 context tokens = 16,384*(32-7-22.9) (confirms the quote)

On Windows and with the MTP variant (MTP heads require a bit more space), the context tokens are going to be even less than what is in the quote.

Agentic workflows often exceed 60,000 context tokens (so, forget about running them).

In this memory sense, any gaming laptop that has a 6-8 GB VRAM GPU is superior and probably even quite a bit cheaper. It's also going to have upgradable RAM.

Recalculating with an additional 8 GB VRAM GPU:
165,500 = 16,384*(32-7-22.9+8).

And, generally, you get more memory per buck if you get a desktop PC.

[1] artificialanalysis.ai/?models=qwen3-8-27b%2Cqwen3-8-27b-medium%2Cqwen3-8-27b-low%2Cqwen3-8-27b-non-reasoning%2Cqwen3-6-35b-a3b%2Cqwen3-6-35b-a3b-non-reasoning&intelligence=artificial-analysis-intelligence-index
[2] reddit.com/r/Qwen_AI/comments/1vo8pjz/qwen3827b_kv_cache_works_out_to_64_kibtoken_so/

[32GB RAM not for local AI]

News from today:

AA-AgentPerf-Local: Benchmarking local AI agents on laptops and workstations

So running LLM locally is becmoing a bigger and bigger thing, and now even ArtificiAlanalysis is posting about it. (but of course the 32 GB RAM AMD version wont be suitable for bigger and hence smarter AI models)