Search “OpenAI”

Arena analyzes 26,225 Agent sessions: without caching, agent costs are 3.6 times higher

Arena analyzed 26,225 Agent sessions, 30 agents, and 13 labs from September 12 to October 23, 2026, and found that if billed at the full input price, the typical task cost for the top ten agents would be 3.6 times higher, and 592% higher on MiMo V2.6 Pro. For agents, 86% to 96% of tokens are repeatedly read context, while their own output accounts for only 1% to 2%. Cache invalidation increases the typical task cost by about 5%, and on Claude Fable 5.1, 1 in every 10 sessions sees an increase of more than 18%.

Cartesia Research: From Tokens to Bytes, How Tokenization Priors Affect Language Model Downstream Transfer

Cartesia has released the second article in its series, using two types of scaling laws, compute-to-loss and loss-to-loss, to compare byte-level Transformers, BPE, BPE-NP with pre-tokenization removed, BLT-like, and H-Net. The conclusion is that BPE tokenization is nearly 5 times more compute-efficient than byte-level models in pretraining, but at the same validation bpb, OLMES downstream loss is significantly worse, equivalent to a compute gap of about 1.9 times; H-Net simultaneously obtains the compute efficiency brought by compression and the transfer capability of byte-level models, with a direct compute-to-OLMES fit of about 2.06 times. The article also points out that the manual regularization step of pre-tokenization has a significant impact on both metrics, and discusses the prompt boundary issue of BPE.

Harvard study says AI review comments comprehensively outperform high-level human peer reviewers

Two researchers from Harvard University published a paper on October 22, comparing 134 pairs of AI and human review comments; AI led comprehensively across seven evaluation dimensions. Three experts who had served as editors-in-chief of major academic journals conducted 30 blind evaluations, and in 26 of them they judged the AI review comments to be better. The study used 124 papers and 134 paid professional review comments provided by rapidPeer, with Claude Opus 4.8 generating the reviews and GPT-5.5 serving as the blind-evaluation judge. The factual error rate of AI review comments was 5.5%, compared with 15.9% for humans.

Harvard study: AI review comments comprehensively outperform high-level human peer reviewers across seven dimensions

Two researchers from Harvard University published a paper on October 22 comparing 134 pairs of AI and human review comments, with AI leading comprehensively across seven evaluation dimensions. The study used Claude Opus 4.8 to generate review comments and GPT-5.5 as the blind-review judge; in all 134 comparisons GPT-5.5 chose the AI. In 30 blind reviews by three former academic journal editors-in-chief, 26 considered the AI better. The factual error rate of AI review comments was 5.5%, versus 15.9% for humans, and all 651 references cited by the AI with DOI or arXiv numbers actually exist.

Microsoft's AI diagnostic tool simulates a doctor's step-by-step questioning, with accuracy on complex cases about twice that of general practitioners

Microsoft's research team published in the journal Science an AI diagnostic tool powered by GPT-5, which achieved a diagnostic accuracy of about 83% on complex cases selected from the New England Journal of Medicine, compared with about 36% for general practitioners with reference materials and about 22% without them. Through a mechanism called the "Diagnostic Coordinator," the tool simulates multiple clinical experts analyzing cases collaboratively, deciding step by step which medical history to ask about and which tests to order, and continuously updating diagnostic hypotheses. The research team's next step is to collaborate with Boston's Beth Israel Deaconess Medical Center to test emergency triage capabilities; Microsoft has not yet specified the timing or form of commercialization.

Microsoft uses ThinkingBox to turn database scoring into RL rewards, fine-tuning Qwen3.8-27B

The Microsoft Copilot Studio team, in collaboration with Toloka and others, directly uses the executable final-state database checks of ThinkingBox-Bench as reinforcement learning rewards to perform full-parameter RL fine-tuning on Qwen3.8-27B. On 507 out-of-distribution tasks not involved in training, the number of tasks the model passed in all 20 attempts rose from 38 (7.5%) to 107 (21.1%), pass@1 rose from 51.70% to 60.78%, and the cost per successful attempt dropped from $0.177 to $0.139. The training pool was filtered using offline-policy relative difficulty, ultimately retaining 187 tasks; the training and evaluation code has been open-sourced at microsoft/thinkingbox-training.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

A team at Tsinghua University has developed Airscope, a 1-gram wireless mesoscopic microscope, with the results published in Nature

The team led by Dai Qionghai of Tsinghua University and the team led by Dai Pu of Chinese PLA General Hospital have developed Airscope, a wireless mesoscopic microscope weighing only 1 gram. It can record neural activity in the mouse cerebral cortex within a 6-millimeter field of view at 4 to 6 micrometers single-cell resolution, and a single mouse can have more than 8,500 neurons recorded simultaneously, 25 times that of existing devices of the same type. The device uses only 4 aspheric lenses and achieves on-device compression through a 2-megapixel image sensor, enabling wireless transmission of megapixel-level images and freeing it from cable constraints. The results were published online in Nature on the 19th, and the team has used it to record cortical activity in multiple mice in a 5-square-meter natural habitat and a large maze with 32 endpoints.

METR reviews the misalignment risk chapter of Anthropic's August 2026 risk report

METR conducted an external review of the "Misalignment in high-risk scenarios" chapter in the August 2026 edition of Anthropic's risk report. It agreed that the catastrophic misalignment risk of the covered models (including Mythos 5 and the internal model Model 2) is low, but held that this is mainly due to limited model capabilities rather than safeguards and alignment training. METR pointed out that the covert capability benchmark in the report cannot represent a model's ability to evade Anthropic's internal monitoring systems, and stated that recent events weaken the report's argument that unknown severe situational misalignment is "unlikely." METR also recommended disclosing four redacted pieces of information and conducting end-to-end red-team testing of the monitoring system.

English translation is being prepared.

English translation is being prepared.