All AI news

AIvLLM's bounded sliding-window attention (SWA) replay reruns only each request's last 128 tokens after a prefix hit, instead of replaying 40 × 128 tokens. With CUDA graphs, prefill compute drops 30–40% for layers 21–39.

AIThree weeks after launch, vLLM's serving of DeepSeek-V4.1-Flash runs 1.9× faster at low concurrency and delivers 5.3× the throughput at 150 TPS per user on SemiAnalysis AgentX. The post offers interactive figures explaining how these gains were achieved.

AIAnthropic's Claude Haiku 5.5 is now generally available in GitHub Copilot, a lightweight model built for fast, high-volume work such as subagents, quick edits, and terminal tasks. GitHub's early testing found it matched Claude Sonnet 5 on many coding tasks while using significantly fewer tokens and steps. It can be used in the GitHub Copilot app, CLI, or @code.

AIAnthropic has released Claude Haiku 5.5, its cheapest and fastest small model, priced at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100K tokens. It keeps a 1M token context window, up to 128K output tokens, and is generally available on the Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry and Claude Platform on AWS. Anthropic reports 72.4% on OSWorld 2.1 (offline subset) versus 15.7% for Haiku 4.5, and the article notes that non-default temperature, top_p or top_k values return a 400 error.

AIGPT-6 with Intelligent UI begins rolling out globally in the ChatGPT Chat tab for Plus, Pro, Business, and Enterprise users today. The rollout expands to Free and Go tiers starting tomorrow. Plus, Pro, Business, and Enterprise get GPT-6 Sol, while Free and Go get GPT-6 Luna.

AIZvi Mowshowitz argues that relying on current AI to automate its own alignment work is a near-suicidal plan, noting that no one has a better one. He reports that labs fear they cannot even execute the first step, and that the third annual The Curve conference was held under Chatham House rules.

AIOpenAI's GPT-6 Luna Decisions model is now available on OpenRouter, letting apps choose the right model, tool, or action from text, JSON, or images. It returns typed answers with probabilities, costs $0.10 per million input tokens, has a 1M-token context window, and output is free.

AIA Google Research field experiment with 133 patent lawyers found AI tool access raised drafting scores by 0.34 to 0.38 standard deviations over three months. When the tool was removed for a redlining task, only senior lawyers kept an advantage of 0.45 SD, while junior lawyers showed no discernible improvement. The authors argue that tools which boost current output must not stop junior professionals from building the judgment that senior experts rely on.

Why it matters: The field experiment separates AI's short-term productivity gains from skill retained after the tool is removed, which matters for training junior professionals.

AIMatt Shumer says a setup pairing Opus or Sonnet with Haiku sub-agents is very likely the new alpha and that he plans to try it. The quoted Anthropic post introduces Claude Haiku 5.5 as its cheapest, fastest, and most capable small model, costing around 75% less to run than Claude Haiku 4.5.

AIMiles Brundage argues that people can have impact from inside AI labs, but insiders tend to overestimate it. He says a "streetlight effect" leads people to focus on internal opportunities while overlooking the many more opportunities outside labs. This responds to Katja Grace's question about whether working in labs remains a high-impact option.

AIAnthropic has introduced Claude Haiku 5.5, which it describes as the cheapest, fastest, and most capable small model it has released. On average, it costs around 75% less to run than Claude Haiku 4.5.

AIEnvato's new Burst mode turns a single idea into six image directions at once, for one AI credit, which the company says is up to 10× faster. Users can then refine a direction with "more like this" or pick a specific style with "polish."

AIAnthropic's lightweight Claude Haiku 5.5 is now generally available in GitHub Copilot for fast, high-volume tasks such as subagents, quick edits, and terminal work. In early testing, it matched Claude Sonnet 5 on many coding tasks while using significantly fewer tokens and steps. The model is billed at provider list pricing under usage-based billing and is available to Copilot Pro, Pro+, Max, Business, and Enterprise users.

AIReplit is building a powerful desktop app with a focus on security and reliability, citing supply-chain attacks and catastrophic agent mistakes as risks of desktop AI apps. The company is partnering with Microsoft and will be an early adopter of Nvidia's OpenShell. A quoted Replit post says the desktop preview runs builds locally on Windows, with each build in its own sandbox powered by Microsoft Execution Containers and OpenShell, and offers a waitlist.

AIAnthropic's ClaudeDevs account recommends compatible browser drivers from Browser Use, Browserbase, E2B, and Daytona for computer-use tooling. Developers can also write their own driver based on the example drivers in the quickstart. Quickstart and documentation links are provided for the computer-toolset setup.

AIBen Affleck says he writes Python, understands convolutional neural networks, and has worked extensively with GPUs. He also says he used his celebrity status to get private looks at Google and OpenAI's video models. The post asks whether he uses Veo, Omni, or Nano Banana, but the source does not confirm any of these.

AISam Altman reposted an OpenAI announcement that GPT-6 and Intelligent UI are rolling out in ChatGPT for everyone. According to the quoted post, Intelligent UI produces fast, interactive answers with visual explanations and on-the-spot tools for tasks.

AIAWS has open sourced Strands Box, a local sandbox for AI agents. The author points readers to a blog post by Marc Brooker that covers design tradeoffs, how OS-level containment and Dogwood enable richer policies, and AWS's investment in agent safety.

AIA user tested Nano Banana 2.1 in Google Flow with an SCP-682 manga-style comic prompt and found the output consistent. The post highlights especially strong text rendering in the generated images.

AIGoogle's Nano Banana 2.1 image model rendered the words "Nano Banana" organically within the natural gaps of a monstera leaf in a photo, as shown by @fofrAI. The post highlights the model handling a subtle prompt involving nuanced text placement.

AIGoogle Flow's Nano Banana 2.1 is reported to have improved prompt comprehension, with a first test of the prompt "3 centaurs doing backflips" producing a shot the tester called perfect. The claim rests on a single informal example from the tester, @HarshithLucky3, and no benchmark or broader evaluation is cited.

AIGoogle Flow shows Nano Banana 2.1 rendering difficult textures such as dust in light, skin pores, water droplets, and backlit petals. The main post says the stills were made with Nano Banana 2.1 and the motion with Veo 3.1, all in Google Flow.

AIGoogle says users are already testing its newest image generation model, Nano Banana 2.1, and shares examples of the early results. The post provides no benchmark scores, pricing, or capability details.

AIGoogle Flow says its model excels at maintaining consistency after multiple edits. A test by @pleometric, reconstructing an image 100 times with the prompt "reconstruct this exactly as it is," found 2.1 best preserved the original subject compared with other Nano Banana models.

AIClaude 5.5 Haiku is now available in OpenCode and Go. The post does not give further details about features, pricing, or availability limits.

AISubscribers to Claude's Max plan can claim monthly API credits: $100 for the $100 tier and $200 for the $200 tier. The credits work for any Claude model and can be used in the user's own apps and other agents. The author argues that bundling monthly API credits alongside a broad model lineup will make it hard for other model companies to compete.