Meta says it will eventually release open-source versions of its new AI models led by Alexandr Wang, but plans to keep certain components proprietary initially.

Microsoft is aiming to develop new large, cutting-edge AI models by 2027 as part of a push to build in-house alternatives to the most powerful AI tools.

Microsoft unveils three proprietary AI models targeting transcription, voice synthesis, and image generation, directly challenging OpenAI and Google.

OpenAI launches GPT-5.4, its most capable frontier model, featuring built-in computer-use capabilities, a 1-million-token context window, 33% fewer hallucinations, and new agentic workflow tools for professional tasks.

OpenAI launches GPT-5.3 Instant, a new model update targeting user experience improvements such as reducing condescending tone, preachy disclaimers, and unsolicited emotional reassurances in ChatGPT responses.

Google releases Gemini 3.1 Pro achieving 77.1% on ARC-AGI-2 benchmark, doubling previous model's reasoning capabilities for complex problem-solving tasks.

Chinese tech giants release multiple AI models before Lunar New Year: Alibaba's Qwen 3.5, ByteDance's Doubao 2.0, Zhipu's GLM-5, and anticipated DeepSeek V4.

Anthropic launches Claude Sonnet 4.6, delivering frontier AI performance in coding, computer use, and agents with 1M token context window, just 12 days after Opus 4.6.

Industry reports suggest Claude Sonnet 5 will match Opus 4.5 capabilities while costing 50% less, with enhanced coding and agent capabilities.

OpenAI discontinues GPT-4o despite emotional user attachments, triggering a #keep4o petition with nearly 21,000 signatures as users mourn the loss of their preferred AI model.

Microsoft develops MAI-1-preview foundation model trained on 15,000 Nvidia H100 GPUs, reducing OpenAI reliance while maintaining partnership through 2032.

Claude Opus 4.6 leads in agentic coding and professional tasks with 1M token context, outperforming Gemini 3 Flash on key benchmarks.

Industry analysis reveals how four frontier AI models—ChatGPT 5.2, Gemini 3, Claude 4.5, and DeepSeek R1—are being strategically deployed in marketing content stacks based on specific workflow requirements rather than benchmark performance alone.

OpenAI announces retirement of beloved GPT-4o and legacy models on Feb 13, 2026, as only 0.1% of users still choose the warm conversational model daily.

Nvidia releases Earth-2 open-source platform with AI weather models for accurate two-week predictions and six-hour nowcasts, advancing climate technology.

In 2026, the focus for enterprise AI is shifting from finding the single "best" model to strategically selecting a portfolio of AI capabilities that align with specific business tasks, risks, and desired outcomes.

Meta's elite Superintelligence Labs has achieved a significant breakthrough with its new AI models, codenamed Project Avocado (text) and Project Mango (visual), with a public release planned for Q1 2026.

At Davos 2026, Nvidia CEO Jensen Huang introduced a five-layer AI infrastructure framework, highlighting key breakthroughs in grounded models, open models like DeepSeek, and the rise of physical intelligence.