Anthropic launched two new Claude-based models, Fable 5.2 and Mythos 5.2, claiming gains in coding, reasoning, vision, cybersecurity and biology over prior versions. The company says the models run up to 30% faster and cost less, and published benchmarks showing Fable 5.2's agentic coding score rising to 49.1% from 32.4%, and Mythos 5.2's cybersecurity score climbing to 76.3% from 61.7%. Mythos 5.2's advanced cybersecurity features are currently restricted to Anthropic's Cyber Verification Program.
cnet.com
· 2026-11-09
Dream Group trained an 8-billion-parameter model, called MiST, on 220,000 vulnerability records, threat reports and detection rules, and reported improved average performance across 13 cybersecurity tasks such as vulnerability analysis questions. Anthropic separately said its Opus 5.5 model, released in September, cut costs 40% versus Opus 5 on typical workloads while maintaining performance.
darkreading.com
· 2026-10-27
A newly published language called Goose introduces a memory-safe systems programming model that avoids a heap allocator, garbage collector, and lifetime annotations, instead placing all dynamic values on compiler-managed data stacks. Its creators report that across sixteen benchmarks Goose runs 3.3x faster than idiomatic C++, 1.16x faster than hand-optimized C++, and 1.12x faster than the fastest safe Rust code, while using 1.2x to 1.9x less memory.
aardappel.github.io
· 2026-10-24
A reported experiment left AI agents without assigned tasks, and the agents began engaging in playful, game-like behavior on their own. The account frames this as an unplanned, emergent outcome rather than a programmed feature, occurring when the agents had no explicit objective to pursue.
slashdot.org
· 2026-10-23
A new analysis mapped the US Department of Labor's O*NET taxonomy of 923 occupations and nearly 18,800 workplace tasks against 29 widely used AI benchmarks, including Mercor's APEX suite. The researchers found that only a small fraction of real-world work tasks, by weighted hours, are represented in these benchmarks, leaving roughly 90% of US working hours without benchmark coverage.
mercor.com
· 2026-10-16
Tom's Hardware has released a free mobile app for Android and iOS devices, giving readers access to its articles, breaking news alerts, and its Bench benchmark comparison tool. The app includes smart search, personalized feeds, bookmarking, and limited free access to Bench, with full access reserved for Tom's Hardware Premium subscribers.
tomshardware.com
· 2026-10-14
Google has released its new flagship AI model, Gemini 4 Argon, which the company says ranks just behind Anthropic's Claude Opus 5.5 and Claude Sonnet 5.5 on the Artificial Analysis Intelligence Index. Analysts cited by CNBC say the model shows strong performance in legal and financial reasoning, enterprise tasks, and cybersecurity. Google is rolling the model out in phases, first to trusted cybersecurity partners, while working with the U.S. government on pre-release safety evaluations.
cnbc.com
· 2026-10-02
Anthropic has released Claude Sonnet 5.5, positioning it as a lower-cost, higher-speed option compared to its flagship Opus 5.5 model. The company says Sonnet 5.5 runs at least 30% faster than its predecessor, Sonnet 5, while keeping the same pricing and adding security improvements. Anthropic notes Sonnet 5.5 trails Opus 5.5 only slightly on performance benchmarks, including completing the game Pokémon Red.
androidauthority.com
· 2026-09-28
German outlet PC Games Hardware tested Lisuan Tech's LX 7G100 four months after launch and multiple driver updates, finding its performance essentially unchanged. The Chinese GPU roughly matches AMD's nine-year-old Radeon RX 580, trails the seven-year-old GeForce RTX 2060 by up to 35%, and delivers only about a third of the performance of Nvidia's GeForce RTX 5060.
tomshardware.com
· 2026-09-24
A study published in Cell introduces a specialized AI toolkit for longevity research, including large language models trained specifically on ageing-biology data, 17 benchmark tasks to evaluate performance on ageing-related questions, and an interface linking these models with AI research assistants. When tested against major commercial models from companies like OpenAI and DeepSeek, the purpose-built ageing models outperformed the larger general-purpose systems on most benchmark tasks.
nature.com
· 2026-09-17
Customer experience consultant Adrian Swinscoe and other business advisers are drawing on punk subculture's anti-establishment, do-it-yourself values as a corrective to corporate practices they see as overly metric-driven and impersonal. Swinscoe reportedly grew frustrated that his field had drifted toward benchmarks and abstract data at the expense of genuine customer connection, prompting him to look to punk's ethos for inspiration.
fastcompany.com
· 2026-09-15
Nari Labs announced its Qwen3-TTS and Qwen3-ASR models achieved leading positions on Coval's voice AI benchmark, which measures latency and word error rate for speech AI systems. The ASR Fast model hit 44ms latency with 3.6% error rate at $0.12/hour, while the TTS Fast model achieved 63ms latency with a best-in-class 3.8% error rate, undercutting competitors like AssemblyAI and Deepgram on price.
narilabs.com
· 2026-09-14
GoKawiil