https://en.wikipedia.org/wiki/Payment_Services_Directive tldr: eu mandates secure 2fa for online banking and most banks only implement it using their app as second factor.

They only offer pay-as-you-go. The $9/month plan includes $2 credits and then its payg without markup.

If such a backdoor exists, it is probably cryptographically secured to prevent "unauthorized" access. E.g. the xz backdoor was secured like that.

There is only one and for this model you need the one with 128GiB RAM.

which start at $3k [0].

[0] https://frame.work/products/desktop-diy-amd-aimax300/configu...

I have the 128 but for Qwen3.5 122b XS quant you only need 64GB

They also do not allow non-coding usage of their coding plan in their ToS.

They are talking about the chat template and not the system prompt. With current gen models, the system prompt is only part of a larger pretext that is passed to the model at the start of the "chat". The models are trained on a specific chat template with things like tool lists, reasoning budget, special feature flags and the "system prompt" formatted in a certain template.

Just did a quick write-up on how to modify Firefox's AI features after spending some time figuring out how to modify the prompts/commands.

Isn't that just Google trying to dark-pattern you into finally clicking that checkboxes you unchecked during setup?

Alibaba released a whole set of new Qwen 3.5 models including a ~120B and a ~35B MoE.

Nice. 27B looks reasonable too.

First impressions of Qwen3.5 35b are that it's amazing

Amazing compared to local models or? How would you compare it to, say, Sonnet 4.5 and Qwen3 Coder Next?

I'm guessing it's fast due to 3B experts

Maybe look into model finetuning/distilation. Unsloth [1] has great guides and provides everything you need to get started on Google Colab for free. [1] https://unsloth.ai/