Tagged articles
model safety
11 articles · Page 1 of 1
The article details how the OrcaRouter team applied abliteration to Qwen3.8‑27B, removed the model's refusal direction, re‑quantized it to FP8, benchmarked unchanged general abilities but near‑zero refusal rates, and provides download links and deployment instructions for red‑team and safety research.
AbliterationFP8Qwen3.8
0 likes · 10 min read

OpenAI is reportedly preparing a new long‑horizon model called Astra, positioned alongside Sol, Terra and Luna, with enhanced multi‑agent coordination and safety concerns that have already sparked internal testing, regulatory review, and widespread speculation about its size, capabilities, and release timeline.
AI agentsAstraOpenAI
0 likes · 7 min read

Anthropic’s Claude Fable 5, released on June 9, is the first publicly available Mythos‑level model that outperforms competitors across code, reasoning, and visual benchmarks, demonstrates autonomous long‑run operation, powers real‑world cases like Stripe’s massive code migration, and introduces a controversial safety‑degradation system.
AI benchmarksAI collaborationClaude Fable 5
0 likes · 11 min read


SuanNiJun 10, 2026 · Artificial Intelligence
Anthropic released two new Claude models—Fable 5, open to all users with a safety classifier, and Mythos 5, a restricted, high‑security version—both achieving record‑breaking performance on software‑engineering, research, vision, and long‑context tasks, while offering a pricing model of $10 per M input tokens and $50 per M output tokens.
AI benchmarksClaude Fable 5Mythos 5
0 likes · 11 min read

Anthropic released Claude Fable 5 and Mythos 5, dual‑version LLMs that achieve record‑breaking benchmarks in software engineering, visual reasoning, long‑context tasks and finance, while introducing a safety‑first routing system, token‑efficiency pricing and a limited free‑trial window, reshaping how developers and enterprises interact with powerful AI agents.
AI benchmarksClaudeFable 5
0 likes · 18 min read

OpenAI’s April 2026 release of GPT-5.5 “Spud” accelerates the weekly‑iteration race among LLMs, and this article dissects its architecture, four major capability gains, benchmark results against Claude Opus 4.7 and Gemini 3.1 Pro, pricing, hallucination risk, safety measures, and advises when to upgrade.
Claude Opus 4.7GPT-5.5Gemini 3.1 Pro
0 likes · 14 min read

Anthropic’s newly unveiled Mythos model surprised its creators by outperforming expectations across more than 100 diverse product‑prototype tests, highlighting emergent capabilities, a strategic shift toward real‑world applicability, and potential implications for AI safety, competition, and industry adoption.
AI competitionAI emergenceAnthropic
0 likes · 6 min read

On February 5, 2026 Anthropic and OpenAI launched Claude Opus 4.6 and GPT‑5.3‑Codex within an hour, sparking a fierce AI model rivalry that brings 1‑million‑token context windows, adaptive reasoning, self‑training, and a shift from AI tools to AI colleagues, reshaping SaaS, developer workflows, and security considerations.
AI agentsClaude Opus 4.6GPT-5.3
0 likes · 13 min read

This report examines DeepSeek’s security and reliability by detailing five core model threats—DDoS attacks, unlimited inference, vulnerability exploitation, data poisoning, and jailbreak—alongside two private‑deployment risks and three external threats such as counterfeit apps, offering targeted mitigation strategies to help users safely adopt the platform.
AI securityDeepSeekmodel safety
0 likes · 8 min read


NewBeeNLPNov 11, 2024 · Artificial Intelligence
This article surveys ten recent multimodal large language model papers, covering vision representation laws, a stricter instruction benchmark, safety impacts of visual adaptation, the Mini‑Gemini architecture, automatic pruning, vision capability boosting, long‑context transfer, efficient token sparsification, math reasoning, and hallucination mitigation.
EfficiencyTraining StrategiesVision-Language Models
0 likes · 18 min read

This article summarizes the Llama 2 series, describing the Ghost Attention technique for maintaining system‑message consistency across multi‑turn dialogs, presenting RLHF and human evaluation results, and discussing extensive safety pre‑training, benchmark assessments, and model release details.
AI-evaluationGhost AttentionLlama 2
0 likes · 20 min read
