GPT-6 Astra led a new 71-task biopharma benchmark with a 70.4% mean score and fully passed 8 tasks, while GPT-6 Sol placed fifth at 54.3% and DeepSeek V4.1 Flash fourth at 57.0%, according to results …
Indirect prompt injection remains a real threat against most models but has dropped sharply on the newest ones, according to Gray Swan's ART and IPI red-team benchmarks: attackers with 15 tries succee…
Xiaomi's MiMo team released MiMo-V2.6-Pro-RL, an open-weight trillion-parameter mixture-of-experts agentic model with 1.02 trillion total parameters, 42 billion active per token, and a 1 million token…
OpenAI unveiled MentalHealthBench on Wednesday, an open benchmark for evaluating model capabilities in mental health domains including safety, seeking user context, preserving user agency, and providi…
A developer built Open Walnut, a self-hosted web app that runs Claude Code sessions, and swapped its LLM-based task-filing guesses for Jev, a decision-only model from TypeSafe AI. On 53 of the develop…
DeepSeek released V4.1 Flash under MIT open weights, priced at $0.003 per 1 million cached tokens, and the model scores 74.2% on DeepSWE, beating Claude Opus 5. The release covers what changed in the …
A developer built a support agent on a Sanity Knowledge Base, deliberately poisoned its help center with seven malicious documents, and found that the Knowledge Base filtered out six of the seven atta…
Developer Angelotc released antigravity-seo, a Go-based SEO and Generative Engine Optimization suite for Google Antigravity that replaces Python/Playwright runtimes with a single Go binary called seo-…
Two independent benchmarks put Anthropic's Claude Opus 5.5 at 37 to 38% of Claude Opus 5's cost per task at each model's default effort setting, beating Anthropic's own claim of "40% less than Opus 5 …
A benchmark of 23,325 judgments across the RAGTruth and SummEval human-labeled datasets found that TypeSafe's Jev matched Claude Opus 5 at 87% accuracy on held-out hallucination detection at roughly 1…
Anthropic published the system card for Claude Opus 5.5, claiming the model matches or beats Fable 5.1 while costing less than Opus 5. The card details five classifier trigger areas with fallback mode…
Anthropic launched Claude Opus 5.5 on September 22, 2026, a model the company says performs comparably to Claude Fable 5.1 on most work while costing 40% less to run than Opus 5 and generating output …
Anthropic released Claude Opus 5.5 on September 22, 2026, the first model in its Claude 5.5 family, which the company says performs at roughly the level of Claude Fable 5.1 on most work while costing …
OpenAI introduced GPT-6 Sol and GPT-6 Luna on September 22, 2026, two models built on methods behind GPT-6 Astra that cut API prices by 50% versus their GPT-5.6 counterparts. GPT-6 Sol is priced at $2…
Anthropic released Claude Opus 5.5, making it available on AWS, Google Cloud, and Microsoft Azure from day one, with Google publishing its own model page for the release. The new model in the Claude 5…
A developer reworked the cost math for Aria, a Go-based text-to-SQL CRM assistant that uses RAG over the database schema, after Anthropic cut Claude Opus 5.5 cache reads to $0.20 per million tokens, 6…
A new arXiv paper (2609.25237v1) proposes a post-training delivery benchmark that tests whether LLM agents can be trusted as forward-deployed engineers (FDEs), rather than whether they can raise a met…
DeepSeek released V4.1-Flash, an open-weight coding model that scores 74.2 on DeepSWE v1.1, essentially matching Claude Opus 5's 74.0 and GPT-5.6 Sol's 73.0 while pricing cached input at $0.003 per mi…
GitHub released Project HydraFusion, a research preview that routes GitHub Copilot CLI coding tasks across multiple models via three execution patterns — Single, Cascade, and Critique — instead of a s…
Every head of evaluations built a personal AI benchmark after Wharton professor Ethan Mollick argued that widely cited tests like MMLU-Pro measure the wrong things, asking models for the approximate c…