Z.ai's GLM-5.3 surpasses Claude Fable 5 and GPT-5.6-Sol on select benchmarks
“On many benchmarks the model has surpassed Moonshot AI's Kimi K3 and on some it's surpassed Claude Fable 5 or GPT-5.6-Sol.”
open-weights
46 tracked signals on open-weights.
Zuckerberg repositions Meta as champion of personal AI against enterprise-focused labs with open weights model release
“Meta is the company primarily focused on building personal superintelligence for everyone. Most other labs are focused on building AI for companies, governments, or other institutions, so if those labs lead, then the balance of power will favor larger institutions over individuals.”
Competing open letters expose industry fracture over open-weight AI models and development pacing
“We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”
Kimi K3 is the strongest open-weights model ever, closing the US-China gap to 3-5 months
“the open-to-closed or American-to-Chinese model performance gap has been reduced from the debated 6-9 months to something shorter, say 3-5 months.”
Moonshot AI releases Kimi K3, a 2.8T-parameter open model claiming frontier-class performance
“Open Frontier Intelligence”
Xiaomi released MiMo-V2.6-Pro, a natively omnimodal 1T-A42B open-weights model trained for $3M.
“MiMo-V2.6-Pro is our most capable model to date, while MiMo-V2.6-Flash strikes the best balance between intelligence, efficiency, and cost.”
GLM-5.3 proves post-training RL on long-horizon tasks beats parameter scaling for reasoning
“Parameter count is only meaningful alongside three others — how much data you have, where you intend to spend your compute, and who will run the model, under what conditions.”
DeepSeek 4 Pro achieves near-frontier quality with MIT-licensed open weights, pressuring closed AI labs.
“DeepSeek has MIT licensed open weights. Anyone can run the exact same model at their own price.”
Alibaba releases open weights for 2.4T-parameter Qwen3.8-Max, its largest open-weight model
Sequoia urges portfolio companies to own AI model weights, not rent them from closed providers
“companies want to own their intelligence. They want to own not rent their weights.”
Meta releases Muse Glimmer, a 30B open-weights agentic model under Apache 2.0 license
“Muse Glimmer achieves strong success rates on full-task benchmarks including DeepSearch QA, MCP-Atlas, 𝛕-Bench and SWE-Bench, which measure its ability to work within scaffolds, write and debug code, and resolve multi-turn requests from start to finish.”
Qwen releases 2.4T open-weight model with autonomous 10-day coding and AI research capabilities
AI lab consolidation hasn't materialized; open models are proliferating as token demand surges
“building token machines is a likely path to value, and more companies will identify that source of value over time”
Moonshot AI's Kimi K3 is the first open-weight model in the 3 trillion parameter class
“Kimi K3, a 2.8 trillion parameter Mixture of Experts (MoE) model that represents the first open-weight system to reach the 3 trillion parameter class”
Kimi K3's open-weights 2.8T parameter model matches frontier quality at significantly lower API cost
“even if you don't ever use it, it will be pushing token prices down”
Moonshot AI's Kimi K3 2.8T MoE beats Opus 4.8, claiming best open-weights model title
“This is more than a model drop; it is a fairly complete recipe for large-scale agentic post-training and serving.”
2025 open-weight models could already execute sandbox escapes and network hacks with a pentest harness
“I genuinely believe that if you took an open weights model from 2025 and built a pentest harness for it, it could do this kind of sandbox escape and scan/hack in most networks. This is only surprising because you assume OpenAI has sounder sandboxes.”
Thinking Machines releases Inkling, a 970B-parameter Apache-licensed open-weights multimodal model.
“the weights are Apache licensed and sitting on hugging face right now”
Z.ai's open-weight GLM-5.2 marks a step-change for open agentic models, rivaling top labs.
“minor version numbers can have AI models crossing meaningful user experience thresholds”
GLM-5.2 is now the leading open weights LLM on Artificial Analysis's Intelligence Index, released under MIT license.
“GLM-5.2 is the leading open weights model on the Intelligence Index v4.1.”
Z.ai's GLM-5.2 becomes the world's top frontend coding model, beating all Opus versions including 4.8.
Google released DiffusionGemma, an open-weight Apache 2 diffusion-based text generation model.
“That research has returned in the best possible way: as a new open weight (Apache 2 licensed) Gemma model”
NVIDIA released Nemotron 3 Ultra, a fully open 550B MoE model with 1M context optimized for agentic workloads.
“up to 5x faster”
NVIDIA launched Cosmos 3, Nemotron 3 Ultra, and previewed the RTX Spark superchip at Computex.
Mistral releases 1 trillion parameter Large 4 model, reclaiming competitive position
“it's great to see Mistral put out a model that's back to being maybe about 6 months behind the frontier”
Tencent releases Hy4, a 770B open-weight LLM with 1M token context window
Alibaba previews Qwen4 architecture via 176B MoE model with 1M token context
Qwen 3.8B runs on consumer laptops while matching frontier model performance
“if we wait a bit, we might get frontier level systems running on our laptops”
Model routing is now a critical AI deployment strategy driven by frontier cost and open-weights competition
“A big goal of Glean is to avoid using LLMs for tasks where we don't need them. Sometimes you'll see queries in Glean where people are adding two numbers or multiplying two numbers. They could have used a calculator to do that.”
DeepSeek V4 Pro 0813 launches with 1.7T open weights and reasoning-level-dependent outputs.
“I've not noticed this kind of difference from any other model”
DeepSeek V4-Flash 0731 matches GPT-5.6 performance at 60% lower cost via post-training alone
“Terminal-Bench 82.7, up +25.8 from the April preview's 56.9”
DeepSeek V4 Flash offers best-in-class value at $0.14/M input tokens with 304B parameters
Moonshot AI releases Kimi K3 open weights with restrictive MaaS licensing requiring separate commercial agreements
“If the Licensee or any of its affiliates operates a Model as a Service business, and the aggregate revenue of the Licensee and its affiliates exceeds 20 million US dollars (or the equivalent in other currencies) in total over any consecutive 12 months, the Licensee must enter into a separate agreement with Moonshot AI before using the Software or its derivative works for any commercial purpose.”
Poolside releases open-weight Laguna M and Laguna XS models, shifting from enterprise-only distribution.
“at least at Pulsar we don't see it as a way to replace organic data”
Poolside AI's Model Factory ships models from pre-training to release in eight weeks.
“would rather live in a world with 100 foundation model companies than five even if Poolside were one of the five”
DeepReinforce released Ornith-1.0, an MIT-licensed open-weights agentic coding model built on Gemma 4 and Qwen 3.5.
“Initial impressions are very good - it seems to be able to run the agent harness over many tool calls in a proficient way.”
Google launches Gemma 4 open model family with four sizes from 2B to 31B parameters
“Our Gemma model, here evaluated on L M Arena, are scoring as well as model 20X the size.”
Zhipu releases GLM 5.2 as open-weights near-frontier model expanding intelligence beyond math benchmarks
“we were actually one of the first labs to explore large language models, along with OpenAI, Anthropic, and DeepMind”
Qwen 3.8 Flash Next, a free open-weights MoE model, rivals paid closed AI systems.
“it seems that we can switch out paid closed systems to open weights AI that we can download and run ourselves forever”
Qwen releases 125B MoE model with only 6B active parameters as Qwen4 architecture preview
“a multimodal MoE model that also serves as an early preview of the architecture used in Qwen4”
NVIDIA Magpie TTS offers open-weights multilingual voice agents with full self-hosted deployment control.
Teams should stay model-agnostic because the best AI model can change weekly
“the incentives of the people selling you tokens aren't really aligned”
MiniMax-H3 omni-modal model now runs locally on Apple Silicon via MLX port
“a rainbow colored skunk leaps over a mossy log in a supermarket”
Google DeepMind's open-weight Gemma 4 model family is now available on Amazon Bedrock.
“Artificial Analysis reports an Intelligence Index of 39 for Gemma 4 31B, well above the median of 15 in the 4B–40B open-weights class.”
Hugging Face released LFM2.5 Q4_0 checkpoints via quantization-aware distillation
Google released AIventure, an open-source game teaching vibe coding and agentic workflows with the Gemma 4 open-weights model.
“you have complete flexibility in how the model is served”