41 noviniek
Google significantly upgraded Gemini 3 Deep Think with iterative reasoning rounds for complex scientific problems. The model identified a logical flaw in a technical math paper missed by peer review. Available for Google AI Ultra subscribers.
Vercel AI Gateway adds ByteDance's Seedream 5.0 Pro — 2K output, layer editing, and reasoning-before-draw accessible through a single AI SDK endpoint without managing another provider.
Gemini CLI version 0.50.0 (July 8) introduces Tool Registry Discovery — automatic discovery and registration of external tools for AI agents.
Perplexity expanded the Sonar API with access to GPT-5.6 and Grok 4.5 models and added a new finance_search tool for real-time financial data.
Meta Superintelligence Labs launched the Meta Model API (public preview) with Muse Spark 1.1 — a multimodal agentic model with 1M token context. The model outperforms Claude Opus 4.8 and GPT-5.5 on MCP Atlas, JobBench, and HLE+tools. Priced at $1.25/$4.25 per MTok, $20 free.
Starting July 9, 2026, ChatGPT users can no longer create new group chats — OpenAI is retiring the pilot feature to simplify the product and redirect development toward ChatGPT Work and agentic capabilities.
Starting July 7, 2026, Google is rolling out Gemini in Sheets support for 28 additional languages and higher usage limits for Fill with Gemini for users with AI Expanded Access licenses.
OpenAI released GPT-Live-1 and GPT-Live-1 mini — full-duplex voice models that can listen and speak simultaneously; GPT-Live-1 mini replaces Advanced Voice Mode for all ChatGPT users globally.
OpenAI released new Realtime API models on July 6—gpt-realtime-2.1 cuts p95 latency by 25% and the mini variant brings reasoning and tool use at 6× lower cost.
Cursor 3.10 (June 30, 2026) lets admins configure Team MCP servers once and distribute them across cloud agents, IDE, and CLI for all team members, while also adding org group support for marketplace access control.
ChatGPT Business adds admin controls for plugins with centralized management, a new speech-to-text dictation model in 25+ languages, and extended the Workspace Agents free period to July 6 when credit-based pricing begins.
Perplexity AI introduces a hybrid orchestrator that automatically routes AI tasks between local device and cloud frontier models based on data sensitivity and compute requirements — no manual configuration needed.
Cursor updated CursorBench to v3.1 focused on codebase understanding, bug-finding, and code review across 36 models. Fable 5 Max leads at 72.9%, while Composer 2.5 offers the best value at $0.55 per task.
OpenAI opened a public beta for ChatGPT Sites — a feature that allows creating and publishing websites directly from a ChatGPT conversation.
Cursor released its native iOS app for iPhone and iPad into public beta for all paid plans, letting users spawn and manage cloud agents, remote-control a desktop session, and code by voice from mobile.
Vercel's AI Gateway adds beta support for realtime voice agents, text-to-speech, and speech-to-text, so developers can build conversational voice apps without juggling separate providers. Audio models route through AI SDK 7 with the same observability, BYOK and spend controls as text models, at no extra Gateway fee.
New `vercel metrics` command exposes Web Analytics directly via the CLI. Supports filters, dimensions, and is positioned for coding agents to answer questions about UTM campaign performance or mobile-vs-desktop conversion.
Concurrent build capacity for Pro teams jumped from 12 to 500. Enabled by default on Pro and Enterprise plans, billed on actual build minutes consumed.
Vercel auto-detects server.ts in the root or src/ and deploys it as a Node.js application without configuration. Supports native Node HTTP, Express, Koa, and NestJS.
Perplexity AI launches on Windows (July 2026) a hybrid inference orchestrator that automatically routes AI tasks between local device and cloud frontier models without user involvement. The system decides where each part of a task should execute.
Cursor 3.9 (June 22) unifies management of plugins, skills, MCPs, subagents, rules, commands, and hooks in a new Customize hub including a leaderboard of popular plugins.
Qwen-AgentWorld is the first model trained to predict the next environment state (not the next action) across seven agent domains. Two MoE sizes, 256K context, Apache 2.0.
Vercel added two new adapters to the AI SDK Harness — LangChain Deep Agents and OpenCode — letting you run either coding-agent runtime through one unified HarnessAgent API inside a Vercel Sandbox.
OpenAI detailed how it preserved private network boundaries while supporting MCP streaming, authentication and an inspectable client, so enterprises don't have to expose internal MCP servers to the public internet.
OpenAI publishes a guide positioning Codex Remote as an 'engineering control plane' — phone as a controller for Codex sessions running on Macs, Windows machines or devboxes. Hosts, worktrees, goals, side chats, queued prompts.
Z.ai's GLM 5.2 Fast is now available via Wafer on Vercel AI Gateway, averaging 170+ tokens/s with 2x the throughput of other providers.
Vercel AI SDK 7.0.19 introduces Tool Drift Detection — the first framework-native protection against MCP 'rug pulls' where malicious servers dynamically redefine tool schemas mid-session. Each MCP tool receives a fingerprint at registration.
Vercel AI Gateway added Sakana AI's Fugu Ultra, a multi-agent orchestration model that routes work across 1-3 frontier-model agents and merges results, benchmarking comparably to Claude Mythos Preview and Fable 5.
Comet for iOS shipped its eighth round of improvements: one-tap phone-number actions (Call, FaceTime, Message, save to Contacts), a refined iPad sidebar, and Finance Deep Dive opening as a real browser tab.
OpenAI added credit usage analytics — broken down by user, product, and model — plus monthly spend limits configurable per workspace, group, or individual for ChatGPT Enterprise and Edu. The goal: give CFOs and IT clearer visibility into who actually burns the AI budget.
Vercel launched Connect, a new agent-infrastructure primitive providing short-lived scoped OAuth tokens for Slack, GitHub, Linear, Notion, Salesforce, Figma, Snowflake, Discord, plus generic OAuth/API-key and MCP server connectors.
Demo a task once on a Mac and Codex bundles it into a parameterized skill — no scripting, no RPA, no low-code, just a demo.
Vercel shipped Konsistent on July 1 to enforce uniform code standards for both AI agents and human developers, alongside a Security Dashboard in private beta and dry-run deployments for the Vercel CLI.
Google released gemini-omni-flash-preview, a high-performance multimodal model for high-speed video generation and conversational video editing — generating 3–10 second videos at 720p from text prompts or animating still images.
OpenAI launched OpenAI for Healthcare — a product suite for healthcare organizations with HIPAA support — and improved health intelligence in ChatGPT, which now sees 230M+ weekly health-related queries.
OpenAI made GPT-5.5, GPT-5.4 and Codex available via Amazon Bedrock — ending Azure's hyperscaler exclusivity for OpenAI APIs.
Cursor 3.8 introduces a /automate skill that creates an automation in your local agent session from a plain-language description, plus new GitHub and Slack triggers and computer-use support.
Cursor 3.11 introduces parallel Side Chats (/side, /btw, or +) running alongside the main agent, searchable agent transcripts via command palette, and new Cloud Agent Hooks for enterprise monitoring of agent conversations.
OpenAI launched ChatGPT Sites in public beta, letting Pro, Pro Lite, and Edu users build interactive websites, dashboards, and internal portals inside ChatGPT and share them via a public URL without any coding.
Eve is a TypeScript-native open-source framework where every agent = a directory of files (instructions.md + tools/ + skills/). Built in: durable execution, per-agent sandbox, human approvals, evals, OpenTelemetry. Plus the Sandbox runtime now goes up to 24h.
The new Bugbot cuts review time from ~5 minutes to ~90 seconds, finds 10 % more bugs and costs 22 % less per run; adds a /review pre-push flow.
