1/11
@Alibaba_Qwen
We're leveling up the game with our latest open-source models, Qwen2.5-1M !
Now supporting a 1 MILLION TOKEN CONTEXT LENGTH
Here's what’s new:
Open Models: Meet Qwen2.5-7B-Instruct-1M & Qwen2.5-14B-Instruct-1M —our first-ever models handling 1M-token contexts!
Lightning-Fast Inference Framework: We’ve fully open-sourced our inference framework based on vLLM , integrated with sparse attention methods. Experience 3x to 7x faster processing for 1M-token inputs!
Tech Deep Dive: Check out our detailed Technical Report for all the juicy details behind the Qwen2.5-1M series!
Technical Report:
https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2.5-1M/Qwen2_5_1M_Technical_Report.pdf
Blog:
Qwen2.5-1M: Deploy Your Own Qwen with Context Length up to 1M Tokens
Experience Qwen2.5-1M live:
Play with Qwen2.5-Turbo supporting 1M tokens in Qwen Chat (
Qwen Chat)
Try it on Huggingface (
Qwen2.5-1M - a Qwen Collection)
Or head over to Modelscope (
Qwen2.5-1M)
2/11
@SexyTechNews
This is why I have millions invested in BABA. Great job, team!
3/11
@TypesDigital
Can you improve the browsing capabilities or access to external links?
4/11
@unwind_ai_
China is getting way ahead with these releases. It feel like somebody just opened a pandoras box.
Boom Boom Boom
5/11
@jacobi_torsten
Great work! But prior Qwen models were barely useful in prior versions for English speaking users! Hope this one is different!!
6/11
@_coopergadd
A million tokens is insane
7/11
@jc_stack
Extended context size is great, but I'm more curious about real-world inference costs at that scale. Love open source models but dealing with memory usage will be interesting.
8/11
@JonathanRoseD
Can we get a Qwen2.5-14B-Instruct-1M but finetuned with Deepseek-R1? Please?
@deepseek_ai
9/11
@anannop
The recurring nightmare of closed-source AI labs.
10/11
@NaturallyDragon
Level up indeed!
11/11
@risphereeditor
This is huge!
To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196