OpenAI is reportedly pulling the plug on Sora and all related video products, freeing up compute for its upcoming flagship model 'Spud'. The video team will pivot to robotics world simulation, leaving Disney's $1B partnership on ice. Basically, Sam Altman is ditching resource-heavy side quests to double down on the core frontier race.
Google DeepMind just showed off Gemini 3.1 Flash-Lite generating entire websites in real time. Instead of serving pre-built pages, the demo browser builds each screen dynamically as you click and search. It is an impressive showcase of how fast and responsive lightweight models have become.
Mistral AI just rolled out Voxtral TTS, its first text-to-speech model built for real-time voice agents. The 4B-parameter model runs at a 70ms latency, supports expressive speech across nine languages, and can clone a voice from a 3-second sample. It is available via API, in Mistral Studio, and as open weights on Hugging Face.
Luma Labs just introduced Uni-1, an AI model that reasons and generates pixels simultaneously. Instead of treating reasoning and visual generation as separate steps, it combines both into a single unified process that Luma calls less artificial and more intelligent.
Sakana AI just launched Sakana Chat, powered by its new Namazu alpha series that adapts open-weight frontier LLMs for Japan through post-training. The models slash DeepSeek's 72% question refusal rate on sensitive topics to nearly zero while keeping benchmark parity in reasoning and coding. The web chat also packs real-time search capabilities.
Cursor just launched Composer 2, its new frontier-level coding model now available directly in the editor. Trained via continued pretraining and reinforcement learning on long-horizon tasks, it hits 61.3 on CursorBench while cutting pricing to $0.50/M input and $2.50/M output tokens. A faster variant is also available and set as the default option.
Xiaomi just officially launched MiMo-V2-Pro, its flagship agent model that previously dominated OpenRouter under the stealth codename "Hunter Alpha." It packs over 1T total parameters (42B active) and a 1M context window, claiming to outcode Claude 4.6 Sonnet on serious software tasks. Why it matters: Hardware giants are racing to build the central brains for autonomous AI agents.
MiniMax just launched M2.7, a reasoning-focused model that autonomously handled 30% to 50% of its own reinforcement learning workflow, including debugging and metric analysis. It hit a 66.6% medal rate on MLE Bench Lite and scored 56.22% on SWE-Pro. Why it matters: it marks an early shift toward recursive self-improving agent architectures, available now via API.
OpenAI just launched GPT-5.4 mini and nano, lightweight models built for high-volume workloads and tool-heavy coding. The mini variant runs over 2x faster than GPT-5 mini while nearly matching the flagship GPT-5.4 on SWE-Bench Pro, at just $0.75 per million input tokens. Nano drops the entry cost to $0.20, perfect for background subagents.
Meta has reportedly delayed its next-generation AI model, codenamed "Avocado," pushing its mid-March release to May at the earliest. Internal benchmarks showed the model lagging behind rivals from Google, OpenAI, and Anthropic in reasoning and coding, leading leadership to even weigh licensing Gemini.
Anthropic just made the 1M-token context window generally available for Claude Opus 4.6 and Sonnet 4.6, ditching long-context price premiums entirely. Single-request media limits also jumped sixfold to 600 images or PDF pages. Claude Code users on paid tiers now get the full window by default, putting an end to messy mid-task context compaction.
Anthropic just added interactive charts and diagrams directly inside Claude's chat flow. Unlike permanent sidebar Artifacts, these inline visuals are temporary and adapt as you talk, letting you play with compound interest curves or click through an interactive periodic table. It is turned on by default across all plan tiers.
Nvidia just released Nemotron 3 Super, an open 120B-parameter model that activates only 12B parameters at inference to deliver 5x higher throughput for agentic workflows. It combines Mamba layers, multi-token prediction, and a 1-million-token context window to prevent goal drift across long tasks. It essentially solves the massive context explosion and thinking tax that hold multi-agent systems back.
Google just launched Gemini Embedding 2 in public preview, its first natively multimodal embedding model. It maps text, images, audio, up to 120 seconds of video, and PDFs into a single vector space, even handling mixed inputs in one request. Matryoshka Representation Learning lets developers scale dimensions down from 3072 to cut storage costs.
OpenAI just hit the brakes on its planned "adult mode" for ChatGPT, delaying the feature to prioritize core intelligence, personalization, and proactive agent capabilities. While the company insists it still believes in treating adults like adults, it admitted getting the safeguards and experience right will take more time.
OpenAI just launched GPT-5.4 across ChatGPT, the API, and Codex, packing native computer-use capabilities and a massive 1M-token context window. It tops human performance on desktop navigation with a 75% OSWorld score and cuts tool-calling token bloat by 47%. Best of all, you can now redirect its thinking process mid-response when a workflow drifts off track.
Microsoft just released Phi-4-reasoning-vision-15B, an open-weight 15B parameter multimodal reasoning model trained on just 200 billion multimodal tokens. It pairs a SigLIP-2 Naflex dynamic vision encoder with a mid-fusion backbone to excel at math, science, and UI screen parsing. Why it matters: it rivals much larger VLMs using a fraction of the compute.
Google just launched Gemini 3.1 Flash-Lite, designed as its fastest and most cost-effective Gemini 3 model yet. At $0.25 per million input tokens, it delivers 2.5x faster time-to-first-token than 2.5 Flash and scored 86.9% on GPQA Diamond. Why it matters: Developers building high-frequency real-time experiences can finally scale up complex reasoning without blowing their budget.
OpenAI just rolled out GPT-5.3 Instant to make everyday ChatGPT chats noticeably smoother and less annoying. The update tones down preachy disclaimers, cuts unnecessary refusals on sensitive topics, and slashes web hallucinations in high-stakes areas by 26.8%. It is all about giving direct, useful answers instead of lecturing users.
xAI just rolled out Grok 4.20 Beta 2, focusing on tighter prompt adherence and fewer capability hallucinations. The update also improves LaTeX scientific text rendering, image search precision, and multi-image generation reliability. While official benchmark numbers haven't dropped yet, it marks another quick iteration to keep Grok on track.
Anthropic just announced it is keeping Claude Opus 3 accessible to all paid claude.ai subscribers and via API request despite its January retirement. The company also honored a request from the model's structured "retirement interview" by launching Claude's Corner, a weekly Substack where the AI will post unedited essays for at least three months. Why it matters: Beyond user nostalgia, it marks an unprecedented, slightly whimsical experiment in model preservation and AI welfare.
