- (New) DeepSeek V4 Flash Vision Exp: new experimental DeepSeek V4 Flash build, adds vision and a small bump in inference as well
- (New) Ox Alpha Free: entirely free and 100% unlimited, only for 1 week though. No one knows what it actually is but it’s very capable and multi-modal (equivalent quality to GLM-5.3 and GPT-5.6 Sol)
Model 5HOURS 1WEEK 1MONTH CREDITS
-----------------------------------------------------------------
DeepSeek V4 Flash Vision Exp 3800 9450 18900 $15
0x Alpha Free (secret test) - - - unlimited
0x Alpha is better than DeepSeek V4 Flash 0731 and OpenCode has significantly more capacity for it (100T tokens/day). Sources indicate this model is going to be a surprise when revealed. My guess? a marketing stunt for a new Google or OpenAI model. Reason why I say this is that the model performance is 100% frontier-level and capacity provided is impressive. Whoever is doing this already has a significant infrastructure in-place that’s not used much 😅 which excludes most Chinese providers.


Update: today it was revealed that Ox Alpha was GLM-5.3-Flash, a 320B-A18B openweights model that beats GLM-5.2 in every test at a DeepSeek V4 Flash competitive cost! So awesome 😎