Full-text search across 61772 articles. Combine with topic and date filters; results sorted by relevance.
Tencent released a preview of a new open-source AI model on Friday, targeting software engineering, research, and financial analysis tasks, and made it available on Hugging Face. The company did not disclose the model's …
An engineer found that AI agents generating short video clips report success on every run, yet roughly one in five outputs were broken—with frozen frames, audio drift, and splice pops that the agent could not detect beca…
Lovable, an AI-driven software development platform, reported a 230% improvement in p50 fetch latency and 160% improvement in push latency after adopting Code.Storage from Pierre Computer Company, which sustained 99.99% …
Microsoft Research's Webwright lets AI web agents write command-line programs instead of clicking through tasks, boosting the same GPT-5.4 model's success rate on long-horizon tasks from 33.5% to 60.1%. The approach leav…
DeepSeek V4 Pro 0813 solves 82.7% of DeepSWE tasks at $8.28 each when escalated to Claude Fable 5 only on failure, outperforming Fable alone (69.7% at $21.63) by 13 points at 62% lower cost, according to a comparison by …
Google released Gemini 3.7 Flash, an AI model with improved reasoning, achieving 43.6% on FrontierCode 1.1 Main (up from 34.4%) and 65.3% on DeepSWE v1.1, available via API and enterprise at $0.75 per 1M input tokens thr…
A developer advises against over-prompting reasoning models, noting that modern models already verify and pace themselves, so extra instructions like 'double-check your work' cause over-verification and wasted tokens. Th…
Emergent, an AI-powered software builder founded by brothers Mukund and Madhav Jha in 2024 with Y Combinator support, has raised $230 million and reached a $1.5 billion valuation, closing a $130 million Series C in July.…
A developer with 50+ customer deployments reports that most AI pipelines are a 'Frankenstack' of Claude, OpenAI, and open-weight models, and recommends routing tasks by context length, output schema, and cost rather than…
DSPy shipped dspy.RLM, a Recursive Language Model module in version 3.1.2, to address context rot — the tendency of models to lose performance on long inputs before reaching their advertised context window size, as ident…
A developer describes how a staged rollout strategy for a shared AI workflow tool prevented a 'fan-out disaster' that would have broken CI across 12 repos. By testing on two opposing pilot repos—one using a pinned packag…
A developer argues that optimizing AI workflows through context pruning, precise prompting, and local pre-processing can reduce the environmental and social costs of large language model infrastructure. The author recomm…
A new open-source framework from University College London (UCL) evaluates open-weight large language models on longitudinal data preparation tasks without sending data to the cloud. Benchmarking models up to 31-35B para…
An ACM survey of 763 computer science educators from 49 countries found that 68 percent have already changed their exams due to AI, shifting toward oral exams, proctored tests, and project-based work, though nearly half …
A new simulation tool built with a Flask backend and JavaScript frontend models how blocking a critical node in a global oil trade network triggers cascading reserve depletion, even in countries with no direct imports fr…
A developer integrated Google's Gemini Omni Flash model into a FastMCP server to create a stateful video generation skill for the Antigravity CLI, enabling iterative refinement via interaction IDs. The system supports fi…
Kimi K3 matches Claude Fable 5 on DeepSWE benchmark quality with a 68.5% pass@1 versus Fable's 69.9%, but costs $4.65 per rollout compared to Fable's $13.41, delivering 2.8x more solved tasks per dollar. The open-weight …
Developers using the Anthropic SDK to access Sakura AI Engine's gpt-oss-120b model must replace the standard api_key parameter with auth_token for Bearer authentication and filter response content by type to avoid crashe…
A vulnerability in macOS Terminal allowed attackers to exfiltrate sensitive data by embedding malicious ANSI escape sequences in text rendered by the terminal, triggering DNS lookups to attacker-controlled domains. Apple…
Cisco released Antares, a family of open-weight AI models with 350 million, 1 billion, and 3 billion parameters, that ranks source files likely to contain vulnerabilities based on a Common Weakness Enumeration descriptio…