Part 5/8:
However, early benchmarks from Artificial Analysis presented a mixed picture. It reported qwq’s performance on the GPT QA Diamond benchmark, at 59.5%, trailing behind Deep Seek R1 and Gemini 2.0 flash. Conversely, it did adhere to its claims in the Amy 2024 benchmark at 78%. Hence, while the model exhibits incredible speed, there’s still room for accuracy improvements with respect to certain tasks.