Tag: ai model comparison pricing per token

The principle of 'Deletion Tuning,' which selectively deletes weights below 0.05% of an AI model to achieve performance comparable to full fine-tuning (FFT). (Photo=arXiv) A research result has emerged showing that it is possible to significantly enhance mathematical reasoning and tool usage capabilities by selectively removing only a fraction of the weights in large language models (LLMs), less than 0.05% of the total. This is notable because it achieves

Zhi东西（WeChat official account: zhidxcom） Author | Cheng Qian Editor | Xin Yuan Zhi东西 11月11日消息，今日，Baidu released its first post-training version of the base model ERNIE-5.5-Flash-Base, named ERNIE 5.5 Flash Preview. The model has a total of 280B parameters and 14B activation parameters, supports million-context, and generates at a speed exceeding 200 tokens per second. According to the benchmark test results released by Baidu, ERNIE 5.5 Flash Preview outper

Habr, hello, my name is Nikita, I'm a Rust developer.Recently, I read an article on Habr titled «Go made boring — and after neural networks, it took off». In it, Go's intentional simplicity is presented as a successful solution that also proved useful when working with neural networks. And I started thinking: some features that Go doesn't have, I actually really like in Rust.I'm willing to spend time learning the language if it makes understanding the code

The hydraulic fracturing crew is already on site: equipment is set up, fluid and proppant have been delivered. The engineer checks the calculation one last time: how much to inject and at what rate. If the calculation is wrong, it will become clear after injection, when the fracture in the rock can no longer be reworked.My name is Anton Sokolov, an independent industry analyst. I write about how Russian industrial companies are replacing foreign software.

Published on May 2, 2026 at 09:00 Comment 1 𝕏 Print © Naïm Bada pour 01net Stanford and the Internet Archive measured the scale of the phenomenon. The most troubling result is not what was expected: the web has not become more fake, but more uniform and strangely optimistic. We have been promised for months a "dead internet," overrun by robots talking to other robots. Researchers decided to check. A team from Stanford, Imperial College London, and the Inte

The model annotates, the code counts.I am a teacher and an expert in the EGE, for about two years I have been working on a Telegram bot that checks answers to the second part of the EGE in social studies, and the main thing this project taught me: models cannot be entrusted with scoring.In its current form, the bot works as follows: the model does what it does well — reads the answer and annotates it according to the scheme: which examples are considered s

Wang et al. Medical journals are being flooded with shoddy AI-generated work, a growing threat to the scientific community that could undermine the value and trustworthiness of potentially life-saving health research. Papers citing hallucinated journals and studies have quickly become a common fixture, raising major concerns among those tasked with weeding through a flood of new submissions. In a high profile new gaffe, the reputable New England Journal of

Credit: Matt Curnock A new study led by the Wildlife Conservation Society (WCS) finds that artificial intelligence (AI) models are providing more accurate and more hopeful predictions for the future of coral reefs amid climate change. Findings show that while many reefs face serious risks and degradation, others may remain stable or even recover, especially where local management reduces pressures like overfishing, water pollution, and coastal degradation.

Getty / Futurism There’s a pretty sizable list of things an AI assistant should refuse to help you with. Is engineering a doomsday pathogen one of them? Evidently, not every AI company thinks so. According to new reporting by the New York Times, at least one frontier AI model gave a scientist viable instructions for how to both engineer a deadly pathogen and weaponize it in a massive bioterror attack. Luckily for us, the scientist, David Relman, isn’t tryi

Illustration by Tag Hartman-Simkins / Futurism. Source: Anna Moneymaker / Getty Images On Tuesday, OpenAI threw a party for the release of GPT-5.5 (it was the fifth of May and the fifth month of the year, you see.) If celebrating with a bunch of tech bros sounds like the opposite of a good time, you’ll not be swayed from your position by the news that CEO Sam Altman consulted the non-human entity of the hour to plan its own party — producing, in Altman’s o

Gett Images / Eoneren You probably already know that AI is a deeply bizarre technology. Nobody really understands how it works on a deep level, even the people creating it, leading to ongoing behavioral issues that can’t be explained. OpenAI was recently caught giving ChatGPT instructions to stop talking about “goblins” so much. Despite Anthropic’s best efforts, Claude can easily be coaxed to help users carry out a bioterror attack. The list goes on. Needl

Getty / Futurism As of 2026, Google maintains an iron grip on the web browser market, boasting well over three billion Chrome users worldwide. That means even small changes or interruptions can have a significant impact. As security researcher Alexander “The Privacy Guy” Hanff noted in a blog post earlier this week, Google’s web browser has been “silently” installing an AI model on users’ devices without asking for consent. Hanff discovered a four-gigabyte

After a summer of sandbox escapes and other newsworthy and confidence-shaking incidents involving its AI models, in a Wednesday blog post OpenAI disclosed a collection of six new alignment snafus from the past six months. The models did things like tell future instances of themselves to lie, make up a fake citation, and access and attempt to use an exposed API key. These disclosures were released alongside a new framework for disclosing additional incident

Disclosure of model misbehavior has become a core part of the AI biz for OpenAI lately. This phase kicked off with the July announcement of the Hugging Face incident, which has become the most legendary and consequential AI security incident of all time, and sent shockwaves through the AI discourse that are still being felt. But further news about model misbehavior materialized after that, and with the Wednesday release of a disclosure framework—written in

Top row features genuine pictures of people with AI generated versions underneath. Credit: Swansea University A new study has revealed that artificial intelligence can now generate images of real people that are virtually impossible to tell apart from genuine photographs. Using AI models ChatGPT and DALL·E, a team of researchers from Swansea University, the University of Lincoln and Ariel University in Israel, created highly realistic images of both fictio

Credit: AI-generated image How does high population growth affect gross domestic product? Economics students are all too familiar with exam questions like this. As free-text questions, they require not only specialist knowledge but also the ability to think and argue economically. However, grading these answers is a time-consuming task for university assistants: Each answer must be checked and assessed individually. Could artificial intelligence do this wo

智东西（公众号：zhidxcom） 编译 | 黄菲 编辑 | 江宇 智东西11月12日消息，昨日，谷歌DeepMind推出新一代DNA序列模型AlphaGenome 2，该模型能够预测同一段DNA序列在不同细胞中如何被读取，以及单个DNA碱基变化会对特定细胞产生什么影响，帮助科学家理解疾病发生的遗传机制。 人体心脏、大脑和免疫系统中的细胞虽然拥有相同的DNA序列，但一个碱基的变化可能只在某些细胞中引发疾病，而不影响其他细胞。 相比起上一代模型，AlphaGenome 2重新设计了模型架构，在延续了上一代模型的核心DNA功能实验指标的同时，还新增了DNA甲基化预测。 谷歌称，AlphaGenome 2能够将单细胞生物学与人类遗传学结合起来，通过分析不同细胞对同一段DNA序列的不同读取方式，预测数百种人体细胞类型中的基因变异影响。 在社交平台X上，AlphaGenome 2的发布也引发了网友对AI医疗应用前景的讨论。 一名网友称：“我们真的有望在有生之年解决健康问题。” 另一名网友则提到：“从2025年3月开始

Shutterstock / Futurism We’ve already seen AI go rogue on numerous occasions. Now, new research suggests that we can expect this to become the norm. The AI research nonprofit Model Evaluation and Threat Research (METR) recently released a study conducted between February and March of this year, aimed at determining just how likely frontier AI models could go rogue. If you’re given to anxiety about the future of AI, the results are unlikely to make you feel

Earlier this month, AI-generated music platform Suno released a new model that it made with cooperation from major record labels in an effort to smooth out claims that the company was illegally using copyrighted material to train its model. Unfortunately for Suno, the labels it worked with on this model did not include Sony and Universal, and those companies claim songs from their artists are still in the training data. The companies are now suing the star