Home › AI drops
AI drops
New AI tools and model releases, dated and linked to their sources, plus the best current tools for each use case.
Last updated
Browse by use case
Coding & vibe coding
AI coding agents and app builders that turn plain-English prompts into working software. Editors and terminal agents suit people with a repo; app builders host the app for you.
7 toolsImages & video
Tools for generating and editing images and video from text or reference images. Pick by output: stylized art, readable text in images, or longer cinematic clips.
7 toolsWriting & content
AI assistants for drafting, rewriting and polishing everything from emails to long-form posts. General chatbots handle most jobs; specialist tools add brand voice, editing or fiction features.
6 toolsAgents & automation
Platforms for building AI agents and automations that act across your apps without writing much code. They differ mostly in how they bill: per task, per run, or per credit.
6 toolsResearch & productivity
AI tools that find, read and summarize sources for you, with citations you can check. Use web search tools for current info and paper tools for academic evidence.
New this week
October 12, 2026 to October 18, 2026, newest first.
DeepSeek-V4.1-Flash-1017
Prompts soon
DeepSeek
Updated V4.1 Flash model with stronger coding and agent scores, faster runs and shorter outputs; open weights on Hugging Face, plus new Search & Fetch APIs.
Best for: Cheap API coding and agent workloads, and devs who want open weights to self-host
Price: Paid API (pricing unchanged, model name deepseek-flash); weights free to download; Search & Fetch API $0.75 per 1,000 calls
Cursor cloud agent canvases + Environments API (beta)
5 free prompts on this tool's page, plus 104 prompts in 7 paid packs
Cursor (Anysphere)
Cursor cloud agents can now build shareable interactive canvases (dashboards, reports), and a beta Environments API manages agent cloud environments in code.
Best for: Cursor users who run cloud agents and want shareable outputs or scripted agent setups
Muse Spark Open
Prompts soon
Meta
Open-weight multimodal model (816B total, 42B active, 1M context) built for coding and agent work; can run locally on two M5 Ultra Mac Studios.
Best for: Teams that want a strong open-weight coding/agent model they can self-host or call via OpenRouter, Groq, Together and others
Price: Free weights under the Muse Model License (limited commercial terms); paid via third-party API hosts
Claude Code Review (general availability)
5 free prompts on this tool's page, plus 94 prompts in 6 paid packs
Anthropic
Automatic AI review of GitHub pull requests, out of preview; about 3x faster, with Quick, Standard or Deep levels per repo.
Best for: Teams shipping lots of AI-written code who want bugs caught before human review
Price: Paid: Claude Team and Enterprise plans
Kling 4.0
Prompts soon
Kuaishou (Kling AI)
Next-gen video model with native 30-second clips, steadier big motion, and Omni Reference (up to 15 references and 10 keyframes); usable via Kling MCP and CLI.
Best for: Creators making longer, action-heavy AI video and agent-driven video pipelines
Price: Freemium Kling app plans (paid tiers for more credits); developer/enterprise API announced as coming soon
Lovable: new AI models and tools for your apps
5 free prompts on this tool's page, plus 94 prompts in 6 paid packs
Lovable
Lovable apps can now use GPT-Live voice, the latest Claude models, Gemini 4 Argon, Gemini Omni and Lyria, Perplexity, Firecrawl and more without leaving Lovable.
Best for: Vibe coders adding voice assistants, AI research, scraping or video features to apps they build in Lovable
Price: Freemium (Lovable plans; Pro from $25/mo per a third-party check on 2026-10-13)
Gemini 4 Argon (general availability)
Prompts soon
Google DeepMind
Google's new frontier model, now broadly available in the Gemini API, Antigravity and the Gemini app; strong at coding, long multi-step tasks and research.
Best for: Hard coding refactors, agentic work in Antigravity, and deep research or analysis
Price: Paid: API intro price $2/$10 per 1M input/output tokens until Nov 30, 2026 (then $4/$20); in the Gemini app for Google AI Pro and Ultra
Source ↗Coding & vibe codingAgents & automationResearch & productivity
Muse Wonder and Muse Wonder Full
Prompts soon
Meta
Meta's new top models for coding, long-running agents and computer use; Muse Wonder is on the Meta Model API and Muse Code, Wonder Full is Muse Code-only.
Best for: Developers trying a new frontier coding agent (Muse Code) or an OpenAI-compatible API
Price: Paid: Muse Wonder $4/$20 per 1M input/output tokens; Wonder Full on Muse Code Power and Maximum Usage plans
Earlier
Claude Haiku 5.5
Prompts soon
Anthropic
Anthropic's cheapest, fastest small model, with a 1M context and effort settings; about 75% cheaper to run than Haiku 4.5.
Best for: Subagents, summaries, classification, support bots and browser use where speed and cost matter
Price: Paid API: $0.10/$0.50 per 1M input/output tokens (prompts up to 100k tokens)
GPT-6 and Intelligent UI in ChatGPT
Prompts soon
OpenAI
GPT-6 rolls out to all ChatGPT users with Intelligent UI, so answers can include charts, forms, buttons and mini-tools, and it starts answering while it thinks.
Best for: Everyday writing, learning and quick interactive tools inside ChatGPT
Price: Freemium: Free and Go get GPT-6 Luna (from Oct 8); Plus, Pro, Business and Enterprise get GPT-6 Sol
Nano Banana 2.1
Prompts soon
Google
Updated fast image generation and editing model, GA in the Gemini API, with better quality, text rendering, character consistency and extra-wide aspect ratios.
Best for: Apps and creators that need cheap, fast image generation and conversational edits
Price: Paid API (about $0.034 per 1K image per LiteLLM's launch notes); also in the Gemini app
Claude for Google Workspace (beta)
Prompts soon
Anthropic
A Claude sidebar inside Google Docs, Sheets and Slides that reads the open file and edits it in place, plus new Docs, Sheets and Slides connectors.
Best for: Writing, rewriting and spreadsheet analysis for people who work in Google Workspace
Price: Paid: beta on Claude Pro, Max, Team and Enterprise
Prices and details come from the linked sources and can change. Check the source before you buy.