xAI pushed Grok 4.3 to the public API on April 30 and bundled Custom Voices, a one-minute voice cloning workflow, at no per-voice surcharge. The release consolidates text, voice, and creative agents on a single xAI bill.
DeepSeek V4 Preview ships MIT-licensed 1.6T and 284B MoE models with 1M context at sub-$0.30 per million output tokens. What it means for creators.
Alibaba released Qwen3.6-27B on April 22, 2026. The 27B dense open-weight model beats its larger 35B-A3B sibling on coding, agent, and vision benchmarks.
Alibaba released Qwen3.6-Max-Preview on April 20, 2026, topping six programming benchmarks including SWE-benchPro. Here is what it means for creators building AI workflows.
Moonshot AI's Kimi K2.6 scores 58.6 on SWE-Bench Pro, placing first above GPT-5.4 and Claude Opus 4.6. Full architecture breakdown, benchmark analysis, and what the open-weight license actually allows.
Arcee AI releases Trinity-Large-Thinking under Apache 2.0, ranking #2 on PinchBench behind Claude Opus 4.6 at just $0.90 per million output tokens.
PrismML emerged from stealth and open-sourced Bonsai 8B, a true 1-bit large language model that fits 8.2 billion parameters into just 1.15 GB of memory, roughly 14x more compressed than comparable 16-bit models.
H Company released Holo3, a vision-language model optimized for GUI agents that sets new state-of-the-art on the OSWorld-Verified benchmark, beating proprietary models with only 3B active parameters.
Anthropic confirmed that Claude paid subscriptions have more than doubled in 2026, with credit card data showing record subscriber growth in January and February.
Xiaomi has launched the MiMo-V2 family, featuring a multimodal model that processes audio, images, and video alongside a TTS engine capable of contextual emotion and singing synthesis.
Anthropic has published the largest multilingual qualitative AI study ever conducted, interviewing 81,000 people across 159 countries in 70 languages about what they want from AI.
MiniMax, the company behind the Hailuo video generation platform, released M2.7 with 10 billion activated parameters and built-in self-improvement capabilities.
Anthropic has launched Dispatch, a new research preview for Claude Cowork that lets users send AI tasks from their phone and have Claude execute them on their Mac desktop.
Mistral launched Forge at NVIDIA GTC, a platform that lets enterprises train frontier-grade AI models from scratch using their own proprietary data.
OpenAI released GPT-5.4 mini and GPT-5.4 nano on March 17, bringing the capabilities of its flagship GPT-5.4 model to significantly smaller, faster, and cheaper packages.
Researchers from CMU, Princeton, Together AI, and Cartesia AI published Mamba-3, a state-space model architecture that matches transformer performance while running inference 7x faster.
NVIDIA launched the Nemotron Coalition at GTC, uniting Black Forest Labs, Cursor, Mistral AI, Perplexity, and four other labs to co-develop open frontier AI models.
Mistral AI released Mistral Small 4, a 119B mixture-of-experts model that unifies reasoning, multimodal vision, and coding under a single Apache 2.0 license.
Anthropic announced on March 13 that its full 1 million token context window is now generally available for Claude Opus 4.6 and Sonnet 4.6 at standard API pricing. The company eliminated the long-context surcharge that previously added up to 100% to the cost of requests exceeding 200,000 tokens
Meta has delayed the launch of its next-generation AI model, codenamed "Avocado," from March to at least May 2026. Internal testing revealed the model falls short of competitors from Google and OpenAI, landing somewhere between Google's Gemini 2.5 and Gemini 3 in performance
NVIDIA released Nemotron 3 Super on March 11, 2026, an open-source 120B-parameter hybrid Mamba-Transformer model that delivers 5x higher throughput than its predecessor for agentic AI workloads
Anthropic launched Claude Code Review on March 9, 2026, a multi-agent system that dispatches teams of AI reviewers on every pull request. The tool uses color-coded severity ratings and boosted thorough code reviews from 16% to 54% in internal testing
OpenAI launched GPT-5.4 on March 5, 2026, introducing native Computer Use mode that lets the model read screens and control mouse and keyboard inputs directly. The release also brings a 1M token context window, 33% fewer factual errors across benchmarks, and 47% token efficiency gains
OpenAI released GPT-5.3 Instant on March 3, 2026, replacing GPT-5.2 Instant as the default model in ChatGPT. The update reduces hallucinations by 26.8% on high-stakes queries, eliminates the preachy tone users complained about, and ships to all ChatGPT users including the free tier