Why is the best Language Model is tweaking in languages?

<!-- SC_OFF --><div class="md"><p>I got curious about something. People constantly say one model is smarter than another. So I ran an experiment.</p> <p>I took ten common prompts,

<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1t0rvwi/its_now_referencing_deleted_chats_fyi/"> <img alt="It’s now referencing deleted chats, FYI 🙂"

<!-- SC_OFF --><div class="md"><p>I just got this pop-off saying that thinking mode now actually improves the accuracy of the images, and it does... like quite a lot. Suddenly, it

<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1wvjetq/visual_math_of_an_llm/"> <img alt="Visual Math of an LLM"

<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1wibolh/how_high_should_your_coding_model_be/"> <img alt="How high should your coding model be?"

<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1t99ewq/chatgpt_has_been_writing_nouns_in_asian_languages/"> <img alt="ChatGPT has been writing nouns in Asian

<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1ss5wt6/new_image_model_generated_indians_by_default/"> <img alt="New image model generated indians by default"

<table> <tr><td> <a href="https://www.reddit.com/r/OpenAI/comments/1xa60rq/psa_chatgpt_intentionally_swaps_selected_model/"> <img alt="PSA: ChatGPT intentionally swaps selected