Latest AI releases, benchmarks, research papers, and funding rounds.https://kolayvibe.com/enhttps://kolayvibe.com/news/claude-4-5-opus-1m-context/https://kolayvibe.com/news/claude-4-5-opus-1m-context/Anthropic ships its most capable model yet with a million-token context window, native computer-use capabilities, and a new pricing tier targeting enterprise workloads.RELEASEhttps://kolayvibe.com/news/gpt-5-swe-bench-78/https://kolayvibe.com/news/gpt-5-swe-bench-78/OpenAI's flagship surpasses all previous models on the canonical software-engineering benchmark.BENCHhttps://kolayvibe.com/news/deepseek-v4-architecture/https://kolayvibe.com/news/deepseek-v4-architecture/The Chinese lab reveals mixture-of-experts details and training compute breakdowns for DeepSeek-V4.PAPERhttps://kolayvibe.com/news/mistral-640m-series-c/https://kolayvibe.com/news/mistral-640m-series-c/The Paris-based lab will use the funds to expand its API, enterprise sales, and open-weight research.FUNDINGhttps://kolayvibe.com/news/gemini-2-5-flash-2m-context/https://kolayvibe.com/news/gemini-2-5-flash-2m-context/Google expands context parity with Claude and adds Grounding API improvements in the same drop.RELEASEhttps://kolayvibe.com/news/llama-4-scout-open-coding/https://kolayvibe.com/news/llama-4-scout-open-coding/Meta's latest open release beats every prior open model on HumanEval and achieves near-GPT-4 on MBPP.BENCHhttps://kolayvibe.com/news/perplexity-shopping-agents/https://kolayvibe.com/news/perplexity-shopping-agents/The search AI adds agentic checkout flows, price tracking, and return management powered by Sonar.PRODUCThttps://kolayvibe.com/news/chain-of-draft-tokens/https://kolayvibe.com/news/chain-of-draft-tokens/A new prompting technique from MIT and Stanford replaces verbose CoT with terse intermediate steps.RESEARCHhttps://kolayvibe.com/news/cohere-500m-enterprise-rag/https://kolayvibe.com/news/cohere-500m-enterprise-rag/The Toronto-based company plans to double its R&D team and expand into on-premise deployments.FUNDING