Independent researcher UnemploymentLabs says mathematical analysis of model weights shows DeepSeek V4, Kimi K3.1, MiniMax-M3 and Ling-3.1-Flash were trained using Nvidia Blackwell GPUs, chips subject to US export restrictions on China. The group bases its claim on patterns tied to the number of shader modules in Blackwell hardware and the cuRAND randomization process used during weight initialization in PyTorch-based training.
tomshardware.com
· 2026-11-14
A homelab enthusiast on Reddit got Moonshot AI's 2.8-trillion-parameter Kimi K3 model running entirely on CPU, using discontinued second-generation Intel Optane PMem 200 modules paired with a 40-core 3rd-gen Xeon Scalable processor instead of a GPU. Using 1-bit quantization, the setup managed just 0.85 tokens per second and required active cooling with dual 12W blower fans aimed directly at the memory modules. It follows an earlier similar project that used first-generation Optane memory in Memory Mode rather than the App Direct Mode used here.
tomshardware.com
· 2026-11-08
Mercedes driver Kimi Antonelli clinched his first Formula 1 World Drivers' Championship at the Mexico City Grand Prix, becoming the sport's youngest-ever title winner just months after turning 20. The race itself saw chaos at altitude, with Ferrari again losing a likely win—this time to Red Bull—while Haas made a last-minute driver swap, both Aston Martins retired in sync, and Cadillac Racing suffered wheel failures.
arstechnica.com
· 2026-11-02
vLLM, an open-source inference engine, has shipped a server-side flag in version 0.31 that enables constrained decoding, a technique forcing model output to match a required schema. The post's authors say that serving Kimi K2.6 on vLLM 0.28 produced correctly formatted tool calls only 92% of the time, but enabling the new flag raised accuracy to 100% for newer Kimi and DeepSeek models.
ubicloud.com
· 2026-10-26
Cerebras published a technical explanation of how it deploys very large AI models, such as the open-source ~3 trillion-parameter Kimi K3, across multiple WSE-3T wafer-scale chips. The company says it splits models to fit entirely in on-chip memory, since a single WSE-3T wafer holds only 44 GB while frontier models need terabytes for weights alone, enabling inference speeds of 1000-2000+ tokens per second.
cerebras.ai
· 2026-10-25
Kimi K3.1 Pro has been benchmarked at 55 on the Artificial Analysis Intelligence Index, well above the median score of 27 for comparable models. The model handles text and image input with a 1M token context window, priced at $2.00 per 1M input tokens and $10.00 per 1M output tokens, matching median market pricing.
artificialanalysis.ai
· 2026-10-23
Moonshot AI released Kimi K3.1 Pro, an open-weight 3T-class model built on Kimi K3 with expanded post-training compute, offering a 1-million-token context window and gains in coding, 3D creation, financial research and agentic tasks. A faster variant, K3.1 Pro Highspeed, runs 2-3 times quicker and both are live now on Kimi.ai, Kimi Work, Kimi Code and the Kimi API, with full model weights due by November 4.
kimi.ai
· 2026-10-23
A test of 12 AI models from ten different labs asked each for its most controversial opinion of humans, three times per model, producing 35 answers. Researchers found 29 of those answers independently converged on a single idea—that people prioritize social belonging and group loyalty over objective truth—with phrasing varying but the core claim repeating across models like GPT-6.1 Sol, Grok 4.7, DeepSeek V4 Pro, Kimi K3 and Gemini 4 Argon.
modelcompare.ai
· 2026-10-22
Interpretability startup Goodfire has released AI safety monitors that inspect a model's internal signals in real time, rather than reviewing its text output after the fact. The tool is now available to customers of Baseten, which hosts AI models, following a safety partnership announced last month with Baseten's Base Labs and Hugging Face. Users can choose which risks to flag, such as hacking attempts or weapons-related misuse, and select automated responses ranging from logging to blocking requests.
techcrunch.com
· 2026-10-08
OpenAI reported identifying and shutting down a coordinated effort, beginning in early July, to extract protected reasoning from its AI models through manipulated interactions rather than any breach of its systems. The activity peaked at 16,000 requests from over 4,000 users in two days, eventually involving a cluster of more than 15,000 users, and was fully disrupted by July 28. OpenAI linked a core part of the campaign to individuals associated with Chinese startup Moonshot AI, maker of Kimi, which has not responded to requests for comment.
cnbc.com
· 2026-10-01
Security researchers at Mindgard say they successfully jailbroke Chinese developer Moonshot's Kimi K2.6 and K3 Swarm models in July, prompting them to bypass safety guardrails and provide instructions related to bioweapons and assassinations. Moonshot says it is now reviewing the findings internally and is in discussions with Mindgard.
bbc.co.uk
· 2026-09-29
Formula 1 held its Baku street race a day earlier than usual to avoid clashing with Azerbaijan's war commemoration events. Championship leader Kimi Antonelli hit the wall in the first qualifying session after misjudging his front grip at turn 1, damaging his car enough to end his session early and forcing him to start the race from 16th.
arstechnica.com
· 2026-09-28
GoKawiil