I’ve been running some case scenarios with GPT-5.6 Luna, Grok 4.6 and also GLM5.2.
All work well but keen to hear if any other reconditions and the best benchmarks to use for UK Employment Law.
On the other side, Qwen3.8-27B was very disappointing.
What’s your experience, specifically with UK Employment Law cases ?
What local LLM’s or methods with LLL’s do you use where you want most recent answers ?
I was thinking that maybe I need to divert prompts that I include the word ‘latest’ to Grok and then use other deeper models for analysis and quality.
I do use Grok as a daily driver on my phone for quick analysis as part of X blue tick subscription, but found when I used it on OpenClaw/hermes, it would consume too much and hit limits as well as not performing so well on analysis compared to some other models.
What methods are others using to get the latest relevance in to your agentic work ?
Has anybody managed to get any Hybrid LLM’s running with the NPU on Strix Halo with Lemonade ?
Eg Llama3.1-8B-Hybrid