BanBTP Model Family

Model Base Architecture Parameters Context Status Notes
BanBTP 124M GPT-2 (from scratch) 124M 1024 Archaic Baseline, useless
BanBTP 783M DialoGPT 783M 1024 Archaic Repeats and breaks
BanBTP V2 Gemma 4 31B 1M Obsolete Heavy — needs 200GB+ RAM. Inconsistent reasoning.
BanBTP V3 Gemma 4 5B 1M Current flagship Stable, remembers facts, thinks through problems
BanBTP V4 (coder) Gemma 4 5B 1M Public variant Fine-tuned for coding + reasoning only
BanBTP INFINITE Custom architecture (distilled from V3 + V4) Unknown 2M Open source / Flagship Combines chat + coding. Beats V2 easily.

largest doesn't always mean better.

Thanks to: the Gemma team for creating the Gemma4 models! <3

BanBTP V3 and V4 are built as the same models. Only difference is V4 is finetuned for coding but not so good with BanBTPV3's security system. so go with V3 for overall chatting and reasoning and use V4 for coding purposes as BanBTP V3 is good for any task but not so much with coding. BanBTP V4 is a coding variant of V3. For BanBTP V2, you must have a BEEFY computer. It tho is not very good with its security system as V3 and V4 are and hence is easy to jailbreak.

V4 does not have a locked down system. it is the unlocked BanBTP V3 if you will. BanBTPV4 based on V3 but simply added coding features to it.

BanBTP infinite is the flagship proprietary model. it is codenamed infinite.

BanBTP infinite can be accessed through the BanBTP website (W.I.P). BanBTP infinite is still in development as of now.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Collections including aaro765/BanBTP-comparison