This feels like confusing a serving-runtime problem for a chip-startup opportunity. Agents do change inference patterns: loops, tool calls, branching, long context, KV reuse, burstiness. But most of that is an inference systems problem: scheduling, routing, KV-cache management, etc. Think Dynamo. By the time a new chip co tapes out + builds a compiler stack + wins cloud distribution, NVIDIA/AMD will likely have baked the obvious hardware-level optimizations into existing platforms.

Follow-up on non-English token-inefficiency with more model-language pairs: - Chinese is cheaper than English on major Chinese models - Gemini and Qwen provide least non-English tax - Anthropic has the highest tax by far; Kimi is next - Hindi is the worst-covered language here, despite its massive speaker base

The non-English tax is real. Sutton's Bitter Lesson, translated across languages and normalized to OpenAI English token count: Hindi: OpenAI 1.37×, Anthropic 3.24× Arabic: OpenAI 1.31×, Anthropic 2.86× Chinese: OpenAI 1.15×, Anthropic 1.71× Claude’s tokenizer charges a much higher linguistic tax.

👋Hi /haɪ/, we're the Tencent Hy /haɪ/ team🐧 Today, we open source Hy3 preview (295B A21B), a leading reasoning and agent model in its size, with great cost efficiency. Give us feedback to help improve Hy3 official! 🤗 📖

🚀 DeepSeek-V4 Preview is officially live & open-sourced! Welcome to the era of cost-effective 1M context length. 🔹 DeepSeek-V4-Pro: 1.6T total / 49B active params. Performance rivaling the world's top closed-source models. 🔹 DeepSeek-V4-Flash: 284B total / 13B active params. Your fast, efficient, and economical choice. Try it now at via Expert Mode / Instant Mode. API is updated & available today! 📄 Tech Report: 🤗 Open Weights: 1/n

Introducing GPT-5.5 A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done. Now available in ChatGPT and Codex.

I like to think real learning happens at the community level, not mainly inside an individual brain or model. What matters is the collective process of producing, refining, selecting, and curating knowledge through research, trial and error, criticism, and feedback. AI progress increasingly works like this too: labs, data companies, users, and models all contribute to a shared ecosystem where better data and evaluations feed future systems. Distillation—whether into a brain or a model—is the easier part. Once a rich knowledge base exists, many different minds or models can absorb it and become capable. The real bottleneck is the civilization-scale process that generates reliable knowledge in the first place.