Meanwhile, the chinese and other open weights models are are killing it. GLM 4.5 is sick. Jamba 1.7 is a great sleeper model for stuff outside coding and STEM. The 32Bs we have like EXAONE and Qwen3 (and finetuned experiments) are mad for 20GB files, and crowding out APIs. There are great little MCP models like Jan too.
Are they AGI? Of course not. They are tools, and that’s what was promised; but the improvements are real.
Turns out, closed source tech bro corporate enshittification is not a sustainable plan. Who knew…
Nah, I tried 20B and a bit of 120B. For the size, they suck, mostly because there’s a high chance they will randomly refuse anything you ask them unless it STEM or Code.
…And there are better models if all you need is STEM and Code.
Look around localllama and AI communities, they’re kinda a laughing stock even more than Llama 4.
Meanwhile, the chinese and other open weights models are are killing it. GLM 4.5 is sick. Jamba 1.7 is a great sleeper model for stuff outside coding and STEM. The 32Bs we have like EXAONE and Qwen3 (and finetuned experiments) are mad for 20GB files, and crowding out APIs. There are great little MCP models like Jan too.
Are they AGI? Of course not. They are tools, and that’s what was promised; but the improvements are real.
Turns out, closed source tech bro corporate enshittification is not a sustainable plan. Who knew…
Did you miss that open ai released the oss model few days prior to gpt5?
Larger model: https://huggingface.co/openai/gpt-oss-120b Smaller model: https://huggingface.co/openai/gpt-oss-20b
They seem to be quite good
Not saying that openai would be the good guys here, but I believe they are realizing that they are behind on local models.
I haven’t played with it too much yet but Qwen 3 seems better than GPt-OSS
Nah, I tried 20B and a bit of 120B. For the size, they suck, mostly because there’s a high chance they will randomly refuse anything you ask them unless it STEM or Code.
…And there are better models if all you need is STEM and Code.
Look around localllama and AI communities, they’re kinda a laughing stock even more than Llama 4.
https://mk.absturztau.be/notes/ab3gv6iygjam02uj
In my limitrd testing it seemed relatively trigger-happy on refusals and the results were not impressive either. Maybe on par with 3.5?
Although it is fast at least.