Jun 7, 04:36 AM

Alibaba Releases Qwen2-72B AI Model with 72B Parameters, Rivals GPT-4 in 15 Benchmarks

Alibaba has released Qwen2-72B, a state-of-the-art open-source AI model with 72 billion parameters and 128K token support. This model rivals OpenAI's GPT-4 and has demonstrated superior performance in 15 benchmarks, including language understanding and generation, with an MMLU of 84.32. Qwen2-72B is now available on the low-bit quantized open LLM leaderboard and supports multilingual capabilities. The model's release marks a significant advancement in AI technology, positioning Alibaba as a formidable competitor in the AI space against giants like Meta and OpenAI. The model has already processed 44M tokens with a throughput of 95+ tokens per second.

#Alibaba #OpenAI #Meta

Written with ChatGPT (GPT-4o).

Sources

OpenRouter@OpenRouterAI
2 years ago
The new @Alibaba_Qwen 2 72B is now available, with top-tier performance among open source models. Already 44M tokens, and 95+ tokens per second throughput. Try it here: https://t.co/30fjB5zsrr
AInnovation@AInnovationAI
2 years ago
New Qwen2 AI model from @AlibabaGroup to take on giants like @Meta and @OpenAI! 🤖💥 Exciting times ahead for AI innovation. Are you ready to see what this game-changer has in store? #AI #Tech #Innovation https://t.co/MRmzkD5hfo
PicoCreator - AI builder @ SF 🌉@picocreator
2 years ago
So while Silicon Valley tries to block open source AI China moved ahead with - qwen2 72B, sliding pass Facebook llama3 70B with clean Apache2 license - kling, launches a sora like model, before sora Like they really dun care about your fears or world view, and accelerate

Additional media

Image #1 for story alibaba-releases-qwen2-72b-ai-model-72b-parameters-rivals-gpt-4-15-benchmarks

Alibaba Releases Qwen2-72B AI Model with 72B Parameters, Rivals GPT-4 in 15 Benchmarks

Sources

Additional media

Similar Stories