The update introduces Exa Lite at $4 per 1,000 requests and Dynamic Highlights that use 95% fewer tokens than full page text.
The model achieves 73.7% on DeepSWE v1.1 and 82% on CyberGym, with output speeds reaching 200+ tokens per second.
The update clarifies that plan-included credits expire at the end of the subscription term while purchased credits expire one year after purchase, and grants UK and EU consumers the right to cancel free of charge if service changes have a more than minor negative impact.
The new backend enables eager execution and native distributed APIs in Transformers, Diffusers, and Kernels, replacing the previous Optimum Neuron workflow.
The model uses diffusion for parallel text refinement and supports a 1M-token context window for latency-sensitive workloads.
The inaugural assessment highlights Gemini's ability to deploy agents across enterprise surfaces with built-in governance and access to over 200 models.
The method addresses specific biases like verbosity and position bias, where LLM judges can flip verdicts or miss errors up to 75% of the time.
Google Cloud and RemotiveLabs enable full virtual vehicle development on Arm-based infrastructure, using RemotiveTopology to simulate the full electrical and electronic architecture.
The open-weight multimodal model features a 1M-token context window and improved performance in coding and agentic tasks.
Tested on Gemini 3.5 Flash and GLM-5, the system improves performance on legal and office benchmarks without changing model weights.
The entire editor, including timeline and effects, is built from extensions that the agent can modify in real time.
The system uses a compiler to resolve code structure, freeing AI cognition for higher-level decisions at a cost of $12.96.
The new HybridAgents framework dynamically switches between latent and text reasoning, improving performance by 10.4% across 11 benchmarks while cutting costs by 2.7x-3.7x.
Three new 'Aurora' model references briefly appeared in OpenAI's API and are being tested on Arena, with observers speculating they may be GPT-Image-3, though this remains a guess.
The fund's stake can increase by an additional 2% at the board's discretion for every 50% rise in company value, bringing current public-good equity commitments to roughly 20%.
Requiring two later sessions to reproduce a skill's change before reuse eliminated poison exposure in tests with Claude Haiku 5.5 and Sonnet 5.5.
Legal experts warn that despite removing the word 'train', Meta's updated terms allow scraping of websites associated with business accounts for any 'advertising and business products', potentially superseding robots.txt preferences.
The model achieves 11.43% WER on LibriSpeech test-clean with just 1.06 MB of INT8 weights.
Anthropic's proposed AI safety bill would designate the Commerce Department as the agency empowered to halt an AI model, but only if catastrophic harm thresholds of 50 deaths or $1 billion in damages are met.
Sphinx pauses git pushes to ask developers questions about the logic of AI-generated diffs, blocking pushes until the developer demonstrates understanding.
An open-source decision model trained on 2,676 contrastive data points that flip the correct answer by changing just one fact
An API for agents and RAG that charges only when an HTTP 200 response is returned along with actual results.
The new version eliminates repetition loops found in the preview, achieves 0/50 loops on SWE-bench tasks, and introduces an open-source engine enabling 262K context within a 16 GB memory budget.
On macOS, background crash dumps accumulate and consume disk space in Codex Desktop even while the app is functioning normally.
AI Briefing