The inconvenient truth might be that the other models score higher than OLMO because they aren't restricted to purely "open and accessible" training data. Who knows what private or ethically dubious data went into training Mistral or llama, for example.
Yeah. Reasoning models like r1 tend to be good for architecting changes and less optimal for actually writing code. So this allows the best of both worlds.
Apple's marketing team must be so good if they can convince HN users that an iPad is a suitable replacement for a "real" computer. No one who knows what a terminal or root access even means should be fooled by this.