✦PickAIWise
Browse all AI →
VERIFIED RELEASE TRACKER

What changed in AI?
Only the updates that matter.

Track new versions and meaningful feature changes from leading AI providers. Every update links to an official announcement and explains who it is for.

Latest AI releasesMajor models and meaningful feature changes, newest first
06 Aug 202601
OpenAIChat & ReasoningMajor update

GPT-5.6 Sol in ChatGPT

GPT-5.6 Sol in ChatGPT now gives more focused answers, improves factual reliability and adds adjustable reasoning. Free users receive GPT-5.6 Luna as the default model with unlimited text chats.

✓ More focused answers✓ Improved factual reliability✓ Adjustable reasoning slider✓ GPT-5.6 Luna for Free users
BEST FOR

Everyday questions, complex reasoning, professional work and users who want control over reasoning depth.

SOURCEOpenAI official announcementRead official release ↗
31 Jul 202602
ByteDance SeedVideo GenerationNew model

Seedance 2.5

Seedance 2.5 targets longer, more complete video storytelling. It generates up to 30 seconds of audio and video in one pass, with multi-round extension, smoother transitions and stronger reference and editing control.

✓ Up to 30-second generation✓ Joint audio-video output✓ Multi-round extension✓ Stronger reference control
BEST FOR

Short films, advertising, continuous scenes and creators who need synchronized audio-video generation.

SOURCEByteDance Seed official announcementRead official release ↗
24 Jul 202603
AnthropicAgents & CodingNew flagship

Claude Opus 5

Claude Opus 5 is Anthropic's new flagship, improving long-running agents, advanced coding and professional knowledge work while staying at the same price level as its predecessor.

✓ Long-running agents✓ Advanced software engineering✓ Professional knowledge work✓ Configurable effort levels
BEST FOR

Complex codebases, long research workflows, professional analysis and tasks requiring sustained autonomous execution.

SOURCEAnthropic official announcementRead official release ↗
24 Jul 202604
MidjourneyImage GenerationDefault model

Midjourney V8.2

V8.2 is now Midjourney's default model, focusing on aesthetics, image quality and Personalization, with a stronger understanding of each user's visual preferences.

✓ Improved image quality✓ Bolder aesthetics✓ Stronger Personalization✓ Better preference profiles
BEST FOR

Users creating distinctive visuals, style explorations, concept art and commercial creative assets.

SOURCEMidjourney official announcementRead official release ↗
Jul 202605
OpenAIVoice & RealtimeNew model

GPT-Live

GPT-Live is designed for real-time voice interaction. Its July update adds SynthID watermarking to supported generated audio and enables provenance verification.

✓ Realtime voice interaction✓ Low-latency audio✓ SynthID watermarking✓ API provenance checks
BEST FOR

Voice assistants, live customer service, speaking practice and apps that need to verify AI-generated audio.

SOURCEOpenAI official announcementRead official release ↗
30 Jun 202606
AnthropicAgents & CodingNew model

Claude Sonnet 5

Sonnet 5 brings stronger agent capabilities to Anthropic's balanced tier. It can plan tasks, use browsers and terminals, and complete work with less step-by-step guidance.

✓ Autonomous planning✓ Browser and terminal use✓ Stronger agentic coding✓ Balanced cost
BEST FOR

Coding, automation, everyday professional work and teams seeking cost-effective agent capabilities.

SOURCEAnthropic official announcementRead official release ↗
May 202607
GoogleMultimodal & AgentsNew model

Gemini 3.5 Flash

Gemini 3.5 Flash combines fast responses with stronger agentic execution, coding and multimodal understanding for high-volume, tool-driven workflows.

✓ Fast agentic execution✓ Stronger coding✓ Multimodal understanding✓ Action-oriented workflows
BEST FOR

High-volume API use, coding agents, multimodal analysis and products where speed is critical.

SOURCEGoogle official announcementRead official release ↗
05 Feb 202608
KuaishouVideo & ImageModel family

Kling AI 3.0

The Kling AI 3.0 family includes Video 3.0, Video 3.0 Omni, Image 3.0 and Image 3.0 Omni, with greater narrative control and stronger character and scene consistency.

✓ Video 3.0 + Omni✓ Image 3.0 + Omni✓ Greater narrative control✓ Stronger consistency
BEST FOR

Narrative shorts, character-consistent video and combined image-video creation.

SOURCEKuaishou official announcementRead official release ↗
✦

Pricing changes also flow into the directory

A model launch does not mean every plan receives immediate access. PickAIWise verifies model features, plan tiers, free allowances and extra credits separately.

Compare AI plans →