16 items from the week, featuring GPT-5.6: Frontier intelligence that scales with your ambition; GPT-5.6 Sol, along with Terra and Luna, will launch publicly this Thursday; @OpenAI: Sol, Terra, and Luna, our GPT‑5.6 family of models, are starting to roll out now.

2 models from the week, featuring @OpenAI: Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model; @AnthropicAI: Claude Fable 5 will be available again globally tomorrow. After a series of pro; DSpark: Speculative decoding accelerates LLM inference [pdf].

16 items from the week, featuring @OpenAI: Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model; DSpark: Speculative decoding accelerates LLM inference [pdf]; @Reuters: 🔊 Anthropic says the US government is allowing it to release its powerful Claude.

4 models, 1 paper from the week, featuring Noam Shazeer Joins OpenAI; Fearless Concurrency on the GPU: Safe GPU inference in Rust, competitive with vLLM/SGLang [R]; American Government Takes Down Claude Fable.

1 paper from the week, featuring Statement on the US government directive to suspend access to Fable 5 and Mythos 5AnnouncementsJun 12, 2026The US government has issued an export control directive to suspend all access to Fable 5 and Mythos 5.; Claude Fable 5 and Mythos 5: The System Card; @ArtificialAnlys: Claude Fable 5 launched today at #1 on the Artificial Analysis Intelligence Inde.

4 models, 4 papers from the week, featuring Gemma 4 12B: A unified, encoder-free multimodal model; Transformers are inherently succinct; Thousand Token Wood: shipping a multi-agent economy on a 3B model.

1 model, 6 papers, 2 tools from the week, featuring Anthropic raises $65B in Series H funding at $965B post-money valuation; Show HN: Tiny-vLLM, high performance LLM inference engine in C++ and CUDA; "Unified Neural Scaling Laws" paper release [R].

10 models, 2 papers from the week, featuring nvidia-nemotron-labs-diffusion; CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs; Full Attention Strikes Back: Transferring Full Attention into Sparse within Hundred Training Steps.

4 models, 6 papers from the week, featuring Luce DFlash + PFlash on AMD Strix Halo: Qwen3.6-27B at 2.23x decode and 3.05x prefill vs llama.cpp HIP; Routers Learn the Geometry of Their Experts: Geometric Coupling in Sparse Mixture-of-Experts; Neural Weight Norm = Kolmogorov Complexity.

4 models, 3 papers from the week, featuring AlphaEvolve: How our Gemini-powered coding agent is scaling impact across fields; Anthropic researchers detail “model spec midtraining”, which adds a stage between pretraining and fine-tuning to improve generalization from alignment training; Nemotron 3 Super (NVIDIA).

4 models, 3 papers from the week, featuring AlphaEvolve: How our Gemini-powered coding agent is scaling impact across fields; Anthropic researchers detail “model spec midtraining”, which adds a stage between pretraining and fine-tuning to improve generalization from alignment training; Nemotron 3 Super (NVIDIA).

6 models, 3 papers from the week, featuring Nemotron 3 Super (NVIDIA); GPT-5.5 Instant System Card; Agents can now create Cloudflare accounts, buy domains, and deploy.

6 models, 3 papers from the week, featuring GPT-5.5 Instant System Card; How OpenAI delivers low-latency voice AI at scale; Supercharging LLM inference on Google TPUs: Achieving 3X speedups with diffusion-style speculative decoding- Google Developers Blog.