China AI

AIXiaomi MiMo and the TileRT team say a 1T model exceeds 1,000 tps on a single standard 8-GPU node using general-purpose GPUs.

Why it matters: The post shows how FP4 quantization, DFlash speculative decoding, and TileRT kernels combine to reach over 1,000 tps on a standard GPU node.

AIXiaomi MiMo and TileRT released MiMo-V2.5-Pro-UltraSpeed, which the post says reaches output speeds above 1,000 tokens/s on a 1 trillion parameter MoE model.

Why it matters: The post specifies the hardware setup behind the claimed speed, which matters for judging whether the approach can be replicated on standard GPU nodes.

AIByteDance has open-sourced the 1.3B-parameter weights of its Bernini Renderer (Bernini-R), available on Hugging Face as ByteDance/Bernini-R-1.3B-Diffusers.

AIXiaomi MiMo has released MiMo-V2.5-Pro-FP4-DFlash, the FP4 backbone behind MiMo-V2.5-Pro-UltraSpeed, with MXFP4 quantization applied only to the MoE experts and a BF16 DFlash drafter for block-diffusion speculative decoding.

AIXiaomi and TileRT released MiMo-V2.5-Pro-UltraSpeed, reporting decode speeds above 1000 tokens/s on a 1-trillion-parameter model using a single standard 8-GPU node.

Why it matters: The post traces how FP4 quantization, DFlash speculative decoding, and TileRT kernels combine to reach 1000+ tokens/s on a single 8-GPU node, which is useful for teams weighing inference throughput.

AIMercedes-AMG F1 and Kimi won the Monaco Grand Prix, and Microsoft credited preparation done well before lights out for the result. The post gives no further details on the race or the work involved.

AIThe 8th BAAI Conference will be held June 12–13 at the Zhongguancun Innovation Center in Beijing, with Turing Award laureates and China's large-model leaders attending.

AIBaidu CoBuddy is now available for free on Novita AI, a code-focused model aimed at developers and AI agents.

AIXiaomi MiMo will credit a free Token Plan of the same tier to users whose plan expired before May 27.

AIMiniMax published MiniMax-M3-MXFP8, an MXFP8 quantized variant of its native multimodal M3 model with 1M context, about 428B total parameters and about 23B activated parameters.

Why it matters: The release pairs sparse attention for 1M-token contexts with reported prefill and decode speedups over M2, useful for judging long-context serving costs.

AIMiniMax has released MiniMax-M3, a native multimodal model with a 1M-token context window, roughly 428B total parameters, and about 23B activated parameters.

Why it matters: The source gives concrete attention-efficiency figures and three reasoning modes, which helps readers judge long-context cost against deployment choices.

AIByteDance has open-sourced the inference code and model weights of the Bernini Renderer (Bernini-R), a DiT-based renderer paired with an MLLM-based semantic planner for video generation and editing.

AIPaddleOCR and ERNIE Image are now available as official Dify plugins, bringing document parsing and image generation into Dify's agent workflows.

AIMiniMax released M3, an open-weight model with a 1M-token context window, native image and video input, and desktop operation support.

Why it matters: The post pairs a new sparse attention design with benchmark figures and a 1M-token context window, letting readers judge the architecture's practical effect on long-context work.

AIXiaomi MiMo has published a blog detailing full-pipeline inference optimizations for the MiMo-V2.5 series.

AIXiaomi describes an end-to-end inference optimization for the MiMo-V2.5 series, centered on Hybrid SWA, which it says cuts KVCache storage to roughly 1/7 of Full Attention.

Why it matters: The post explains how Hybrid SWA's theoretical KVCache savings were realized in production through dual pools, SWA-aware prefix caching, and tiered storage, giving concrete engineering patterns for long-context inference.

AIZack Williams built NabuOCR with PaddleOCR to help read cuneiform from tablet images, and it won the ERNIE AI Developer Challenge.

AIJeremy Howard says Xiaomi's MiMo v2.5 model appears to offer very good value, with per-prompt cost shown after each answer. His test context includes more than 40k tokens of tool descriptions, a system prompt, and skills.

AIMiniMax has upgraded its Agent, renamed Mavis, and introduced Agent Teams that run multiple role-based Agents in parallel on desktop.

Why it matters: The post explains why multi-Agent helps long tasks and where its verification, token, and aggregation costs come from, useful for judging when a team setup beats a single Agent.

AIMiniMax says its M2 series could not output the name Ma Jiaqi, a failure it traced to post-training data that rarely included the token.

Why it matters: The post traces a community-noticed token failure through tokenizer, embedding, and lm_head tests, showing how post-training data coverage can cause low-frequency token drift.

AIFunAudioLLM has released Fun-ASR-Nano-2512-hf, a Hugging Face Transformers-compatible version of its end-to-end speech recognition model, which supports Chinese, English, and Japanese.

AIDeepSeek has made its discount on DeepSeek-V4-Pro permanent, after previously extending it until May 31, 2026, 15:59 UTC. The company is encouraging developers to build with the model.

AIYann LeCun points to Project Tapestry, hosted by thealliance.ai, as the answer to fears that no credible Western open frontier model will emerge.

AISebastian Raschka reviews recent open-weight LLM architecture changes aimed at reducing long-context memory and compute costs.

AIBaidu's PaddlePaddle account announced ERNIE 5.1, which it says cuts total parameters to about one-third and activated parameters to about one-half, using roughly 6% of the pretraining cost of models at similar scale.

Why it matters: The post pairs parameter and pretraining cost reductions with benchmark results against named frontier models, letting readers weigh efficiency against capability.

AI🎉 Quoted teaser exception applies: the main post is only a reaction, so the quoted post carries the news.

AIDeepSeek has extended the 75% discount on its DeepSeek-V4-Pro API until May 31, 2026, 15:59 UTC, up from the previously announced May 5 deadline.

AIChina's Ministry of Industry and Information Technology issued the Reference Guide to the Application of Data Elements in Industrial Scenarios through a notice dated April 20, 2026, and published on April 28.

AIXiaomi's MiMo-V2.5 is a native omnimodal model that understands text, image, video, and audio within one architecture.

Why it matters: The repository documents a 310B-parameter omnimodal MoE with a hybrid attention design, useful for comparing long-context efficiency against other open multimodal models.

AIXiaomi released and open-sourced MiMo-V2.5-Pro, a 1.02T-parameter Mixture-of-Experts model with 42B active parameters and a 1M-token context window.

Why it matters: The release pairs a 1.02T-parameter open-weight model with long-horizon agent results and token-efficiency claims, useful for judging its fit in coding and agent workflows.

AIDeepSeek has reduced the price of input cache hits across its entire API series to one-tenth of the original price, effective immediately.

AIDeepSeek is offering a 75% discount on the DeepSeek-V4-Pro API until May 5, 2026, 15:59 UTC. For Claude Code, users can set the model to deepseek-v4-pro[1m] to unlock a 1M context window, while OpenCode requires v1.14.24 or later and OpenClaw requires v2026.4.24 or later.

AIMax Woolf says DeepSeek 3.2 was a very horny LLM, and DeepSeek V4 is even more so. The post offers an informal impression without benchmarks, test details, or examples.

AIAhmad Al-Dahle argues that the most interesting part of DeepSeek-V4 is its bet on efficient ultra-long context rather than its benchmarks.

Why it matters: The post argues that efficient 1M-token context, not benchmark scores, is the key bet behind DeepSeek-V4's design for test-time scaling and long-horizon agents.

AIThe DeepSeek API now supports V4-Pro and V4-Flash through both the OpenAI ChatCompletions and Anthropic interfaces.

Why it matters: The source gives exact model names, an unchanged base URL, and a July 2026 discontinuation date, so developers can plan their migration from legacy names.

AIXiaomi released MiMo-V2.5, a 310B-parameter sparse MoE model with 15B active parameters that adds native visual and audio understanding.

Why it matters: The release pairs native visual and audio understanding with a 1M-token context window and open weights, a combination worth checking against your own multimodal workflows.

AIMoonshot AI released Kimi K2.6, an open-source native multimodal agentic model with 1T total and 32B activated parameters and a 256K context length.

Why it matters: The model card gives specific agent swarm scale, context length, and benchmark comparisons against several frontier models, useful for judging its coding and agent capabilities.

AIOllama and oMLX showed impressive demos running Qwen 3.5 and Gemma 4 locally on Apple silicon using MLX. Awni Hannun said the capabilities of local LLMs and their surrounding ecosystem have advanced considerably over the past couple of years.

AIMiniMax has released MiniMax-M2.7 on Hugging Face, describing it as its first model to participate in its own evolution.

Why it matters: The source ties its benchmark claims to a self-evolution process and a named comparison set, which helps readers weigh how the reported gains were achieved.

AIThe Ministry of Industry and Information Technology released a list of pilot consortia for building high-quality industry datasets for AI empowerment.