ChatGPT-4o Latest and LFM3-1.7B post similar overall scores

OpenAI’s ChatGPT-4o Latest and Liquid AI’s LFM3-1.7B are closely matched on the LLM Stats Score, earning 24.3 and 25.0, respectively. ChatGPT-4o Latest leads the reasoning index with 23.6, compared with 20.8 for its rival. The overall scores draw on uneven evaluation coverage, limiting how directly the results can be compared.

ChatGPT-4o Latest wins the only shared benchmark, GPQA, with a reported score of 84.0%. That head-to-head result provides a narrower measure than the composite indexes and does not establish broad superiority across tasks. LFM3-1.7B has additional reported results covering instruction following, coding and other capabilities, but matching results for OpenAI’s model are not available.

Deployment options also differ. ChatGPT-4o Latest supports multimodal inputs and specifies an input capacity of 128,000 tokens and an output capacity of 16,384 tokens. LFM3-1.7B lacks documented context limits and multimodal support, but offers open weights for self-hosting or fine-tuning under the LFM Open License v1.0. OpenAI’s model is proprietary. Neither model specifies a knowledge cutoff, preventing a comparison of training-data recency.