This small chart tells you why #Mercury2 by #Inception is a big deal and how it is a leap of 11x against #Claude-Haiku4.5 and 14x over #GPT5-mini on real world testing comparisons without hardware upgrades.
#ChatGPT #Anthropic #dLLM #LLM #AI #ML #DiffusionLLM
#ChatGPT #Anthropic #dLLM #LLM #AI #ML #DiffusionLLM
February 25, 2026 at 10:17 PM
This small chart tells you why #Mercury2 by #Inception is a big deal and how it is a leap of 11x against #Claude-Haiku4.5 and 14x over #GPT5-mini on real world testing comparisons without hardware upgrades.
#ChatGPT #Anthropic #dLLM #LLM #AI #ML #DiffusionLLM
#ChatGPT #Anthropic #dLLM #LLM #AI #ML #DiffusionLLM
PUBLIC TALK
with Dr Mubarak Shah, Center for Research in Computer Vision, University of Central Florida.
Explore how diffusion-based AI can become more accurate, robust, and trustworthy.
📅 Mon 18 May, 6–7pm.
📍 Fox Lecture Theatre, UWA
🖱️https://events.humanitix.com/diffusionllm
with Dr Mubarak Shah, Center for Research in Computer Vision, University of Central Florida.
Explore how diffusion-based AI can become more accurate, robust, and trustworthy.
📅 Mon 18 May, 6–7pm.
📍 Fox Lecture Theatre, UWA
🖱️https://events.humanitix.com/diffusionllm
May 15, 2026 at 8:56 AM
PUBLIC TALK
with Dr Mubarak Shah, Center for Research in Computer Vision, University of Central Florida.
Explore how diffusion-based AI can become more accurate, robust, and trustworthy.
📅 Mon 18 May, 6–7pm.
📍 Fox Lecture Theatre, UWA
🖱️https://events.humanitix.com/diffusionllm
with Dr Mubarak Shah, Center for Research in Computer Vision, University of Central Florida.
Explore how diffusion-based AI can become more accurate, robust, and trustworthy.
📅 Mon 18 May, 6–7pm.
📍 Fox Lecture Theatre, UWA
🖱️https://events.humanitix.com/diffusionllm
⚡ Inception Labs: su LLM de difusión es 10 veces más rápido que Claude, ChatGPT y Gemini
Mercury 2, un modelo de lenguaje basado en difusión, promete una velocidad revolucionaria.
https://thenewstack.io/inception-labs-mercury-2-diffusion/
#DiffusionLLM #LargeLanguageModels #AI #RoxsRoss
Mercury 2, un modelo de lenguaje basado en difusión, promete una velocidad revolucionaria.
https://thenewstack.io/inception-labs-mercury-2-diffusion/
#DiffusionLLM #LargeLanguageModels #AI #RoxsRoss
March 3, 2026 at 12:07 AM
⚡ Inception Labs: su LLM de difusión es 10 veces más rápido que Claude, ChatGPT y Gemini
Mercury 2, un modelo de lenguaje basado en difusión, promete una velocidad revolucionaria.
https://thenewstack.io/inception-labs-mercury-2-diffusion/
#DiffusionLLM #LargeLanguageModels #AI #RoxsRoss
Mercury 2, un modelo de lenguaje basado en difusión, promete una velocidad revolucionaria.
https://thenewstack.io/inception-labs-mercury-2-diffusion/
#DiffusionLLM #LargeLanguageModels #AI #RoxsRoss
HEX, a training‑free method that ensembles multiple generation schedules, boosts diffusion LLM accuracy on GSM8K from 24.72% to 88.10%, and improves TruthfulQA to 57.46%. Read more: https://getnews.me/hidden-semi-autoregressive-experts-enhance-diffusion-llm-inference/ #hex #diffusionllm #reasoning
October 8, 2025 at 6:41 AM
HEX, a training‑free method that ensembles multiple generation schedules, boosts diffusion LLM accuracy on GSM8K from 24.72% to 88.10%, and improves TruthfulQA to 57.46%. Read more: https://getnews.me/hidden-semi-autoregressive-experts-enhance-diffusion-llm-inference/ #hex #diffusionllm #reasoning
ParallelBench, the first diffusion‑LLM benchmark, evaluates parallel decoding on tasks like arithmetic and list sorting, showing quality drops despite speed gains. Read more: https://getnews.me/parallelbench-reveals-limits-of-parallel-decoding-in-diffusion-llms/ #parallelbench #diffusionllm
October 8, 2025 at 4:34 AM
ParallelBench, the first diffusion‑LLM benchmark, evaluates parallel decoding on tasks like arithmetic and list sorting, showing quality drops despite speed gains. Read more: https://getnews.me/parallelbench-reveals-limits-of-parallel-decoding-in-diffusion-llms/ #parallelbench #diffusionllm
AGRPO, an on‑policy RL method for diffusion LLMs, improved GSM8K accuracy by up to 7.6% over LLaDA‑8B‑Instruct and gave a 3.8× boost on the Countdown benchmark. Read more: https://getnews.me/new-rl-algorithm-boosts-reasoning-in-diffusion-language-models/ #diffusionllm #agrpo
October 7, 2025 at 9:40 PM
AGRPO, an on‑policy RL method for diffusion LLMs, improved GSM8K accuracy by up to 7.6% over LLaDA‑8B‑Instruct and gave a 3.8× boost on the Countdown benchmark. Read more: https://getnews.me/new-rl-algorithm-boosts-reasoning-in-diffusion-language-models/ #diffusionllm #agrpo
Rainbow Padding cycles through seven distinct padding tokens to curb early termination in diffusion LLMs, achieving length robustness after a single epoch LoRA fine‑tuning. Read more: https://getnews.me/rainbow-padding-improves-length-robustness-in-diffusion-llms/ #rainbowpadding #diffusionllm
October 7, 2025 at 6:08 PM
Rainbow Padding cycles through seven distinct padding tokens to curb early termination in diffusion LLMs, achieving length robustness after a single epoch LoRA fine‑tuning. Read more: https://getnews.me/rainbow-padding-improves-length-robustness-in-diffusion-llms/ #rainbowpadding #diffusionllm
Quant-dLLM introduces a framework for 2‑bit post‑training quantization of diffusion large language models, preserving performance. The code and pretrained models will be open‑sourced on GitHub. https://getnews.me/quant-dllm-2-bit-post-training-compression-for-diffusion-llms/ #quantdllm #diffusionllm
October 7, 2025 at 10:41 AM
Quant-dLLM introduces a framework for 2‑bit post‑training quantization of diffusion large language models, preserving performance. The code and pretrained models will be open‑sourced on GitHub. https://getnews.me/quant-dllm-2-bit-post-training-compression-for-diffusion-llms/ #quantdllm #diffusionllm
SAPO adds step‑level rewards to diffusion language models, aligning each denoising iteration with a hierarchical reasoning plan and boosting benchmark performance. Read more: https://getnews.me/step-aware-policy-optimization-improves-reasoning-in-diffusion-llms/ #diffusionllm #sapoinnovation
October 3, 2025 at 8:53 PM
SAPO adds step‑level rewards to diffusion language models, aligning each denoising iteration with a hierarchical reasoning plan and boosting benchmark performance. Read more: https://getnews.me/step-aware-policy-optimization-improves-reasoning-in-diffusion-llms/ #diffusionllm #sapoinnovation
AdaBlock-dLLM, a training-free scheduler that adjusts diffusion LLM block sizes on the fly, improves accuracy by up to 5.3% while maintaining the same throughput. https://getnews.me/adablock-dllm-adaptive-block-sizing-boosts-diffusion-llm-speed/ #adablockdllm #diffusionllm #semiautoregressive
October 3, 2025 at 12:49 PM
AdaBlock-dLLM, a training-free scheduler that adjusts diffusion LLM block sizes on the fly, improves accuracy by up to 5.3% while maintaining the same throughput. https://getnews.me/adablock-dllm-adaptive-block-sizing-boosts-diffusion-llm-speed/ #adablockdllm #diffusionllm #semiautoregressive
Freedave enables lossless parallel decoding for diffusion LLMs, delivering up to 2.8× higher throughput without accuracy loss, per a paper posted 30 Sep 2025. Read more: https://getnews.me/freedave-enables-lossless-parallel-decoding-for-diffusion-llms/ #freedave #diffusionllm #ai
October 2, 2025 at 8:59 PM
Freedave enables lossless parallel decoding for diffusion LLMs, delivering up to 2.8× higher throughput without accuracy loss, per a paper posted 30 Sep 2025. Read more: https://getnews.me/freedave-enables-lossless-parallel-decoding-for-diffusion-llms/ #freedave #diffusionllm #ai
Learn2PD, a lightweight post‑training filter for diffusion LLMs, boosts decoding speed up to 22.58× (57.51× with KV‑Cache) without quality loss. Read more: https://getnews.me/learn2pd-accelerates-diffusion-llms-with-adaptive-parallel-decoding/ #learn2pd #diffusionllm #paralleldecoding
October 1, 2025 at 4:06 AM
Learn2PD, a lightweight post‑training filter for diffusion LLMs, boosts decoding speed up to 22.58× (57.51× with KV‑Cache) without quality loss. Read more: https://getnews.me/learn2pd-accelerates-diffusion-llms-with-adaptive-parallel-decoding/ #learn2pd #diffusionllm #paralleldecoding
Spiffy speculative decoding speeds diffusion language models by 2.8–3.1×, up to 7.9× with other tricks, while preserving output distribution, according to researchers. Read more: https://getnews.me/spiffy-speculative-decoding-boosts-diffusion-llm-speed-by-up-to-7-9x/ #spiffy #diffusionllm
September 25, 2025 at 7:01 AM
Spiffy speculative decoding speeds diffusion language models by 2.8–3.1×, up to 7.9× with other tricks, while preserving output distribution, according to researchers. Read more: https://getnews.me/spiffy-speculative-decoding-boosts-diffusion-llm-speed-by-up-to-7-9x/ #spiffy #diffusionllm
IGPO adds brief verified reasoning fragments to diffusion language models. Tested on GSM8K, Math500 and AMC, it achieved new state-of-the-art accuracy on these math benchmarks. https://getnews.me/inpainting-guided-policy-optimization-boosts-diffusion-llm-performance/ #inpainting #diffusionllm
September 17, 2025 at 6:39 AM
IGPO adds brief verified reasoning fragments to diffusion language models. Tested on GSM8K, Math500 and AMC, it achieved new state-of-the-art accuracy on these math benchmarks. https://getnews.me/inpainting-guided-policy-optimization-boosts-diffusion-llm-performance/ #inpainting #diffusionllm
Inception Ships Mercury 2 - A Diffusion LLM That Hits 1,009 Tokens Per Second
https://awesomeagents.ai/news/inception-mercury-2-diffusion-reasoning-llm/
#InceptionLabs #Mercury2 #DiffusionLlm
https://awesomeagents.ai/news/inception-mercury-2-diffusion-reasoning-llm/
#InceptionLabs #Mercury2 #DiffusionLlm
Inception Ships Mercury 2 - A Diffusion LLM That Hits 1,009 Tokens Per Second
Inception Labs launches Mercury 2, the first diffusion-based reasoning language model, generating over 1,000 tokens per second on Blackwell GPUs at a fraction of the cost of conventional autoregressive models.
awesomeagents.ai
February 26, 2026 at 3:49 AM
Inception Ships Mercury 2 - A Diffusion LLM That Hits 1,009 Tokens Per Second
https://awesomeagents.ai/news/inception-mercury-2-diffusion-reasoning-llm/
#InceptionLabs #Mercury2 #DiffusionLlm
https://awesomeagents.ai/news/inception-mercury-2-diffusion-reasoning-llm/
#InceptionLabs #Mercury2 #DiffusionLlm
ByteDance just dropped iLLaDA, a diffusion‑based text model that cranks out output 4× faster. Speed’s up, but the MMLU score dips. Curious how this trade‑off reshapes generative AI? Dive in for the full breakdown. #iLLaDA #DiffusionLLM #MMLU
🔗 aidailypost.com/news/bytedan...
🔗 aidailypost.com/news/bytedan...
June 27, 2026 at 10:47 AM
ByteDance just dropped iLLaDA, a diffusion‑based text model that cranks out output 4× faster. Speed’s up, but the MMLU score dips. Curious how this trade‑off reshapes generative AI? Dive in for the full breakdown. #iLLaDA #DiffusionLLM #MMLU
🔗 aidailypost.com/news/bytedan...
🔗 aidailypost.com/news/bytedan...
Imagine running a diffusion LLM straight from your phone’s NPU, thanks to Multi‑Block Speculative Decoding. Faster, private, and power‑efficient AI at your fingertips. Dive into the tech that’s reshaping on‑device generation! #MobileNPU #DiffusionLLM #SpeculativeDecoding
🔗
🔗
June 15, 2026 at 6:06 AM
Imagine running a diffusion LLM straight from your phone’s NPU, thanks to Multi‑Block Speculative Decoding. Faster, private, and power‑efficient AI at your fingertips. Dive into the tech that’s reshaping on‑device generation! #MobileNPU #DiffusionLLM #SpeculativeDecoding
🔗
🔗
FAIR-Calib just dropped a two‑stage PTQ trick that slashes diffusion LLM size without losing quality. Curious how quantization can stay sharp? Dive in for the details! #FAIRCalib #DiffusionLLM #Quantization
🔗 aidailypost.com/news/fair-ca...
🔗 aidailypost.com/news/fair-ca...
June 8, 2026 at 6:04 AM
FAIR-Calib just dropped a two‑stage PTQ trick that slashes diffusion LLM size without losing quality. Curious how quantization can stay sharp? Dive in for the details! #FAIRCalib #DiffusionLLM #Quantization
🔗 aidailypost.com/news/fair-ca...
🔗 aidailypost.com/news/fair-ca...
Inception Labs 推出的 diffusion LLM 實測推論速度不錯。Diffusion LLM 是 autoregressive LLM 以外一個蠻有趣的路線，值得關注後續發展。目前看來 Mercury 2 速度有感提升，可以期待未來在特定應用上的表現。
#DiffusionLLM #InferenceSpeed #LLM
https://x.com/AndrewYNg/status/2026478474681262576
#DiffusionLLM #InferenceSpeed #LLM
https://x.com/AndrewYNg/status/2026478474681262576
May 18, 2026 at 9:15 AM
Inception Labs 推出的 diffusion LLM 實測推論速度不錯。Diffusion LLM 是 autoregressive LLM 以外一個蠻有趣的路線，值得關注後續發展。目前看來 Mercury 2 速度有感提升，可以期待未來在特定應用上的表現。
#DiffusionLLM #InferenceSpeed #LLM
https://x.com/AndrewYNg/status/2026478474681262576
#DiffusionLLM #InferenceSpeed #LLM
https://x.com/AndrewYNg/status/2026478474681262576