TJ-1.0 vs GPT-4o vs Gemini: Tested on Tajik, Russian & English

TJ-1.0 vs GPT-4o vs Gemini: We tested all three on Tajik, Russian & English Hi, I'm Muhammadjon — developer at SoulLab . We built TajikGPT — the first AI platform from Central Asia. One question we get constantly: "How does TJ-1.0 compare to GPT-4o and Gemini?" So we ran honest tests. Here's what we found — including where we lose. Why this matters Central Asia has 15+ million Tajik speakers. No major AI lab has properly trained on Tajik language data. GPT-4o and Gemini treat Tajik as an afterthought — they translate from Russian rather than understanding the language natively. We decided to fix that. Tajik Language Results Task TJ-1.0 GPT-4o Gemini 2.0 Flash Comprehension ✅ Best Acceptable Weak Translation (Tajik ↔ Russian) ✅ Best Partial errors Partial errors Creative writing ✅ Best Generic Generic Instruction following ✅ Best Misses context Misses context GPT-4o and Gemini miss Tajik morphology, cultural context, and regional expressions. TJ-1.0 wins clearly here. Russian & English