Anthropic reported to investors that its quarterly revenue rose nearly 50% to $16.8 bn in Q3, up from $11.5 bn in Q2 and $4.8 bn in Q1. The company highlighted that $1.70 of revenue was earned for every $1 spent on computing power, and its adjusted operating margin (excluding stock‑based compensation) was 11%. It also disclosed that distillation—training models on outputs from more advanced models—accounted for 5‑7% of revenue, a practice it has materially reduced in the third quarter.
The article discusses OpenAI’s focus on Recursive Self‑Improvement (RSI), which the author suggests may represent a new form of AGI. It highlights how OpenAI’s research team is employing coding agents and notes a significant rise in AI spending per researcher in late July, likely linked to internal access to a model later released as GPT‑6 Astra. The piece references related essays and includes a chart illustrating the growth of agentic engineering at OpenAI.
The paper discusses how enterprises increasingly deploy AI coding agent harnesses, often purchased from vendors like Anthropic or OpenAI, and how these harnesses dictate model choice, prompt handling, and cost. It introduces a fast, customizable routing system that classifies prompts and strategically routes them to minimize expensive model usage, achieving 14–21% cost savings in a simulated 10,000-seat enterprise. The study also evaluates risks across twenty harnesses, highlights vendor dependence, and proposes an internal control plane for future harness ownership decisions.
By Arian Abbasi, Alan Aqrawi, Ted Kwartler
E-Commerce Bench is an open‑source benchmark that simulates a year‑long e‑commerce operation, requiring LLM agents to manage multiple online stores, negotiate with suppliers, optimize sales, fulfill orders, handle returns, and manage cash flow. The environment uses real product and supplier data, a calendar of promotions and shocks, and deterministic customer and negotiation models to enable reproducible evaluation. The study evaluates 18 state‑of‑the‑art models across seven metrics, finding no single model dominates, with GPT‑5.6 Sol achieving the highest year‑end assets but lagging in fraud avoidance and operational efficiency.
By Wei Fan, Xinjie Shen, Xudong Guo, Jianhong Tu, Yang Su, Yinger Zhang, Lianghao Deng, Fengyu Wang, Baohua Dong, Yangqiu Song, Dayiheng Liu
GPT‑6 Astra is a new OpenAI model rolling out today to a limited set of organizations and soon to all ChatGPT Plus, Pro, Business, Enterprise users, and via the OpenAI API and AWS. It is priced at $10/million input and $50/million output, matching Claude Fable 5/5.1, and outperforms Fable on most OpenAI self‑reported benchmarks, achieving 99.9% on the ARC‑AGI 3 benchmark with a custom Provider Adapter harness. Astra excels in security tasks—scoring 100% on ExploitBench, 42.4% on ExploitGym, and 99.2% on SRE‑Bench—and handles long context well, hitting 100% on OpenAI’s eight‑needle benchmark at 256K–512K tokens and 96.3% at 512K–1M tokens, though it remains behind Fable on the Intelligence Index and Meta’s Muse Spark 1.3.
Claude Haiku 5.5 is Anthropic’s new fast, low‑cost model that matches GPT‑6 Luna’s pricing of $0.10 per 1,000 input tokens and $0.50 per 1,000 output tokens up to 100,000 tokens, with a 5× increase beyond that threshold. It uses a less generous tokenizer, meaning the same long prompt consumes about 1.25× more tokens than Haiku 4.5, which raises hidden costs. Benchmarks show Haiku 5.5 outperforms Luna within the 100,000‑token limit, but Luna becomes more economical for larger workloads.
arXiv:2608.29843v1 Announce Type: cross
Abstract: Posted prices for AI inference have fallen steadily since 2024, yet the measured speed of that fall depends almost entirely on the method of measurem...
By Louis Yiven Zhu
The paper introduces metrics for measuring the economic value produced by AI systems per unit of energy, namely Net Value per Joule (NVJ) and Time Saving per Joule (TSJ). It evaluates 220 public GDPval tasks across 44 occupations using five open-weight models and four accelerator types, finding that the highest dollar value per task does not always align with the highest value per joule. The study highlights variations in performance across models and hardware, such as GLM 5.2 achieving the highest net dollar value and Gemma 4 31B delivering a higher NVJ and TSJ with less energy.
By Amirreza Zeinali, Jon Saad-Falcon, Avanika Narayan, Mersad Abbasi, Jonathan S. Hartley, John Hennessy, Christopher Re, Azalia Mirhoseini
arXiv:2609.34951v2 Announce Type: replace
Abstract: In 2023, AI models answered from training data and hallucinated when it ran out, and businesses were told to seed that knowledge. Models' training...
By Ido Finder, Assaf Elovic, Gad Shalev, Liad Yosef
arXiv:2606. 18543v1 Announce Type: new Abstract: Language model agents are becoming proficient executors at isolated, short-horizon tasks such as software engineering and customer service.
By Haozhe Chen, Karthik Narasimhan, Zhuang Liu
Analysis provides insights into ChatGPT’s impact on the economy. OpenAI also launches new research collaboration to study AI’s broader effects on the labor market and productivity.
arXiv:2607. 20349v1 Announce Type: cross Abstract: Generative AI can produce book-length works of fiction at near-zero cost.
By Tuhin Chakrabarty, Xinyue Liu, Jane C. Ginsburg, Paramveer Dhillon