John Carmack just pushed back against backlash over Microsoft’s WHAMM AI demo, which generates Quake II frames in real time without a traditional engine. After critics called it a threat to developer jobs, Carmack compared AI to high-level languages replacing hand-written machine code, calling powerful tools central to computing progress. Epic Games CEO Tim Sweeney also voiced support.
GitHub just rolled out Agent mode and MCP support to all VS Code users, turning Copilot into an autonomous developer that edits multiple files, runs terminal commands, and self-heals runtime bugs. It also opened access to Claude 3.7 Sonnet and Gemini 2.0, alongside a new $39-a-month Pro+ plan.
Cognition AI just launched Devin 2.0 in beta, slashing its steep $500 monthly barrier down to a $20 starting plan. The revamp introduces a cloud-hosted IDE capable of running multiple Devins at once, plus interactive planning and automated code documentation. Why it matters: Lowering the price tag gives the high-profile AI software engineer a second chance to prove its worth against surging rivals.
Tencent just rolled out a major upgrade to its Hunyuan T1 deep reasoning model, boosting its project-level code generation and text writing capabilities. The update also sharpens multi-turn comprehension, instruction following, and the handling of mixed Chinese and English text. You can test it out right away on Tencent's platform or Hugging Face.
Windsurf just rolled out its Wave 6 update, bringing the entire app-building lifecycle right into the IDE. In partnership with Netlify, developers can now deploy projects to a live public URL with a single click or Cascade prompt. The release also adds Jupyter Notebook tab support, enterprise controls for MCP, and fixes long-chat context degradation.
OpenAI just rolled out an update to GPT-4o in ChatGPT, sharpening its ability to follow complex prompts with multiple requests and handle tricky coding tasks. The model also gets a bump in intuition and creativity—plus OpenAI deliberately tuned it to spit out fewer emojis.
Anthropic just released its second Economic Index report, analyzing 1 million Claude.ai conversations after launching Claude 3.7 Sonnet. The findings show augmentation still dominates at 57%, with zero occupations fully automated. Extended thinking is heavily concentrated in coding and scientific research, and Anthropic has open-sourced the underlying dataset on Hugging Face.
Anthropic is reportedly preparing to boost Claude 3.7 Sonnet's context window from 200K to 500K tokens, according to newly surfaced feature flags. The jump would let developers drop massive codebases into a single session without leaning on RAG. It remains unclear whether the tier will stay exclusive to enterprise users.
Imandra just launched CodeLogician, an AI coding assistant powered by neurosymbolic AI and automated reasoning. It turns generated code into mathematical models to formally verify behavior and eradicate subtle hallucinations. Python is supported first, with an early-access waitlist open now.
Google just released Gemini 2.5 Pro Experimental, its newest flagship model that reasons through thoughts before answering. The model immediately took the #1 spot on LMArena, boasts a 1-million-token context window, and hits new highs on math and coding benchmarks like SWE-bench. It is available right now in Google AI Studio and for Gemini Advanced subscribers.
DeepSeek just quietly updated its flagship DeepSeek-V3 model, switching to a permissive MIT license that lets developers freely build commercial projects. Community benchmarks show a small boost in coding performance, and an Apple researcher managed to run a quantized version on a Mac Studio at 20 tokens per second. It's a huge win for open-source AI commercialization.
OpenAI just rolled out its compute-heavy o1-pro model to the API for select Tier 1–5 developers. It burns more compute to deliver sharper reasoning responses, while adding vision, function calling, and structured outputs. The catch is the eye-watering price: $150 per million input tokens and $600 per million output tokens.
Razer just unveiled AI QA Copilot, a tool designed to take the friction out of video game testing. As QA testers play through a build, the AI flags crashes and glitches in real time, automatically compiling reproduction steps, logs, and video clips into ready-to-file tickets. Human oversight is still required, but cutting down manual paperwork could noticeably speed up release cycles.
OpenAI just rolled out direct PDF inputs for its Responses and Chat Completions APIs. Developers can now pass raw PDF documents straight to models like o1, gpt-4o, and gpt-4o-mini without third-party parsing tools. Why it matters: document workflows just got a lot simpler.
Open-source tool Browser Use just blew up after users realized it powers the viral AI agent Manus. Co-creator Gregor Zunic says daily downloads jumped from around 5,000 on March 3 to 28,000 on March 10. By letting models click buttons and fill out forms, it is rapidly becoming the shovel of choice in the web agent gold rush.
Nous Research just launched an inference API, giving developers cloud access to Hermes 3 70B and the DeepHermes-3 reasoning model without hosting weights locally. It matches OpenAI's format for drop-in replacement, offers a $5 starter credit, and is rolling out via a first-come waitlist.
Anthropic CEO Dario Amodei predicted that AI will write 90% of software code within three to six months and practically all of it in a year. The bold forecast suggests software engineering will pivot from writing code to reviewing and orchestrating AI-generated logic.
OpenAI just launched a dedicated suite of tools to help developers build AI agents, anchored by the brand-new Responses API alongside built-in web search, file search, and computer use. It also released an Agents SDK and tracing tools to orchestrate complex multi-step workflows. Why it matters: OpenAI is streamlining agentic apps into a single endpoint and plans to sunset the Assistants API by mid-2026.
Researchers just introduced Generalized Interpolating Discrete Diffusion (GIDD) to overcome the rigid limitations of next-token prediction. By blending masking with uniform noise, it achieves top-tier generation performance while finally allowing models to revise and self-correct their own mistakes.
Google just launched an experimental text embedding model, gemini-embedding-exp-03-07, through the Gemini API. Trained directly on Gemini, it took the top spot on the MTEB Multilingual leaderboard with a score of 68.32, beating the runner-up by 5.81 points. Developers can try it out immediately using the existing embed_content endpoint.
