Peaked at #3 Now #3
EAGLE-4 inference release
EAGLE-4 launches, hitting 961.6 tokens/s on Alibaba's Qwen3.8-27B via TokenSpeed on 4 NVIDIA B300s, its fastest yet.
Key points
- EagleCorp released EAGLE-4, which it says is its fastest inference release yet.
- EAGLE-4 reached 961.6 tokens/s on Alibaba_Qwen's Qwen3.8-27B with TokenSpeed, using BF16, 4× NVIDIA B300, TP4, one request, T=0.
- The original poster said one request on Qwen3.8-27B delivers 4× vanilla throughput: EAGLE-4 at 961.6 tokens/s, DFlash 2 at 672.6, DSpark at 618.6, and Vanilla at 240.5, with MegaKernel disabled and TokenSpeed.
- The original poster linked to a GitHub page for EAGLE-4.
Key points and the reaction summary are written by AI from the posts on this page. Check the original post. How we use AI
Original post
Top replies on X
Sign in to see 3 top replies from X
From @EagleCorp, @lightseekorg, @selim_aktas2 and others. Spam removed, with English and Chinese translations.
Discussion 0
Sign up Sign in to join the discussion
No comments yet. Start the conversation.