Unverified50% confidenceBenchmarkExact time
A single RTX 5090 with LLaDA-2.1-mini + ORL reaches 2,788 tokens/s at NVFP4 precision (single-request decoding, batch size 1).
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/17/2026
First Seen
Valid until: 10/31/2026
Sources
Related Entities
Related Claims
UnverifiedOn 8×B200, DiffusionGemma 26B MoE + ORL reaches 7,708 tokens/s at NVFP4 precision (single-request decoding).79% similarUnverifiedIn terms of prompt processing speed, the RTX 4090 achieves approximately 3,000 tokens/s and the M5 Max approximately 800 tokens/s, giving NVIDIA a clear advantage in long-document and codebase input scenarios.72% similarUnverifiedOn an RTX Pro 6000 (96 GB VRAM), the official FP8 version combined with built-in MTP speculative decoding can reach approximately 120 tokens/second, with quality nearly indistinguishable from 16-bit.70% similarUnverifiedOn an RTX 5090, Muse Glimmer runs at 75 tokens/second without speculative decoding, and reaches 233 tokens/second when paired with a speculative decoding model67% similarUnverifiedIn testing, an RTX 5070 running the 2-bit quantized version outputs 94 tokens per second.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1049949API
curl https://kongchang.com/api/v1/knowledge/claims/1049949MCP
get_claim(id=1049949)