← sources

Are any of the other 'free' models worth using?

source: reddit (reddit:LocalLLaMA)published 2026-11-13 22:15ingested 2026-11-14 02:42

retained text

When I was using opencode, I would just use the daily free Deepseek 4.1 allowance for when I wanted to test something using a bigger/better model than my local Qwen-Flash-Next. 

However I'm now using Pi, and unfortunately there's currently no way to use the opencode free models on Pi. And I am yet to find any other provider with free deepseek usage.

I did find a couple ways to get other models, such as the ones below, but most of them I've never tried before and have no idea on their current capabilities. 

Are any of these worth folding into my workflow? 

I guess Qwen 3.8 should be, assuming it's the full model, as I already use a Q5 of that occasionally (but it has limited context of about 128k so the online one would get the full 268k which would be handy too). Depends if full-size Qwen 3.8 is better than the current Q4 Flash-Next I use. 



Dots3-Note Preview (384K)

Nemotron 3 Nano Omni

Nemotron 3 Ultra 550B

Nemotron 3.5 Lightning

North Mini Code

Liquid LFM3 1.7B

Kilo Auto0

OpenRouter 

Glyph Cluster 

Nemotron 3.5 Super VL 

GPT-OSS 120B

GPT-OSS 20B 

Qwen3.5 397B A17B

Qwen3.5 9B

Qwen3.6 27B 

Qwen3.8 27B

Mistral Small 3.2 24B 

Meta Llama 3.3 70B Instruct 

document fetched 2026-11-14 00:46 · canonical https://reddit.com/r/LocalLLaMA/comments/1xvlgkh/are_any_of_the_other_free_models_worth_using