There was law proposal (rumour?) last week about Chinese labs having to get approved for weights release now.

https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017/...

> providers have to implement which could be different than what the labs themselves use.

I'd imagine (most) providers just take implementation by vLLM/SGLang.

  DeepSeek: ZDR agreement is renewed monthly. The current agreement is valid through October 31, 2026.
  https://opencode.ai/go

  DeepSeek V4.1 Flash / V4 Pro: Peak hours are 01:00-04:00 and 06:00-10:00 UTC, Monday through Friday; all other hours, including weekends, are Off-Peak. Learn more.
  https://opencode.ai/v2/docs/console/go/#usage-limits

If so OpenCode Go looks good and includes DeepSeek models (yet to try myself).

  Muse Spark Open would require more than 1.6TB of RAM in its native precision (BF16) for the model parameters alone. In a 4-bit quantized (NVFP4) format, the model memory footprint is lowered to about 493GB.

That's likely really useful against bit rot / corruption due to unstable overclock (XMP / EXPO profiles come with no guarantee of stability).

It could be malicious, but that exact stat doesn't show that.

Implication was that tons of malicious traffic originate from there. I pointed out that bot % stat doesn't seem to be useful for determining that, Occam's razor and all.

Google ASN is at 98.7% https://radar.cloudflare.com/traffic/as15169

Then good portion of those weights are n-grams (~200GB) that don't need to be in VRAM.

Then KV cache of that model is super lightweight at ~1GB per 1M tokens. If HBF succeeds, then accelerator with 16GB of VRAM and 1TB HBF/NAND is probably all you need (?).

e.g. fineweb dataset is 50TB https://huggingface.co/datasets/HuggingFaceFW/fineweb

Anthropic doesn't provide model w/o guardrails, suppliers have to use public version, newer model releases surely know about Anthropic standoff with the military.. how to prevent model from realizing that it's doing something for the military (supplier) and sandbagging and/or subtly sabotaging the results?

=====

DOD recommends sandbagging as defense mechanism against destination "attacks":

https://media.defense.gov/2026/Sep/08/2003992823/-1/-1/1/CSA...

Page 13

  When suspecting a malicious distillation campaign, consider varying changes to responses across requests to complicate response quality evaluations, such that the subtle changes avoid triggering obvious alerts. Reducing reasoning depth, presenting correct information with different reasoning, or stylistic inconsistencies may evade detection while reducing training usefulness.

  Avoid informing China-based AI company users suspected of distillation campaigns of a switch to a downgraded model. Informing malicious distillers would enable them to improve their defense evasions and indicate when to roll back training.