THE CLICK TEST / AI NEWS
AI news, checked against the source.
Clear, dated updates about AI tools and model announcements. We separate what was announced from what people can actually use or test.
Latest update: October 23, 2026
NEW UPDATES · OCTOBER 23, 2026
Latest AI and creator updates
Each report separates company announcements from our own checks. Read the individual update for access limits and what remains untested.
GOOGLE GEMINI · OCTOBER 20, 2026
Gemini adds saved brand styles
Google’s small-business hub brings reusable brand styling and connected tools together. Check rollout, usage limits and Canva restrictions.
Official Google artwork. Saved-style consistency has not been independently tested.
ELEVENLABS · OCTOBER 20, 2026
ElevenLabs adds music and timed effects
Uploaded footage can guide a score and separate effects. Check access and credit limits.
ElevenLabs promotional artwork. Results have not been independently tested.
MESHY · OCTOBER 20, 2026
Meshy edits one part of a 3D model
Selected-area edits cost 20 credits and require supported Meshy models.
Meshy company demonstration. Results have not been independently tested.
ADOBE + NVIDIA · OCTOBER 20, 2026
Frame.io speeds up AI footage search
New processing improvements target large archives. Visual search requires eligible paid access.
NVIDIA promotional artwork. Performance figures are company claims.
FAL + BLACK FOREST LABS · OCTOBER 19, 2026
FLUX 3 Max arrives with faster-video claims
A new paid video model on fal, with introductory pricing. See the access limits and why its speed claims are not our own test results.
Official fal promotional artwork. Proposed sample NOT TESTED.
AFFINITY · OCTOBER 18, 2026
Affinity fixes extra AI images and saving hangs
October build 4940 corrects Seedream’s four-variation default. See the saving fixes and the paid-access limits of Affinity’s older Decompose Layer feature.
Official Affinity/Canva image of the September layer feature. October fixes have not been independently tested.
OPENAI · OCTOBER 16, 2026
ChatGPT Slides: Our Original Editable Test
We created a new workflow slide with editable shapes and connectors, changed its subtitle, and inspected the saved result. Eligible paid Space access is required.
See the actual slide and test limits
Actual slide render created for this article. Exports, collaboration and audio remain untested.
KLING AI · OCTOBER 17, 2026
Kling 4.0: 30-second clips and the rollout catch
Up to ten keyframes and combined references are documented. Some headline output features are still coming soon; access and current pricing need an account check.
Official Kling artwork; no generation test performed.
MICROSOFT PHOTOS · OCTOBER 17, 2026
Photos Relight is expanding—check preview access first
Microsoft announced broader access on October 15. See the Insider preview steps and why Relight may still be missing on your PC.
Official Microsoft demonstration; the new build has not been tested by The Click Test.
WORDCRAFT · OCTOBER 14, 2026
WordCraft’s Free Editor Adds Native Printing
Version 0.7.0 adds printer controls to the open-source Word-style editor. Here is what changed, where to get it and why its alpha status still matters.
Official screenshot: ArtCraft. Release verified; printing not independently tested.
YOUTUBE MUSIC · OCTOBER 14, 2026
YouTube Music Adds AI Labels: What Creators Need to Know
A new player-page label explains when music was made primarily with generative AI. Here is what the rollout means for creators—and what YouTube says about disclosure and monetization.
Illustration: YouTube. Announcement verified; live account rollout not tested.
Intelligent UICHATGPT ROLLOUT
ChatGPT can answer with interactive experiences
OpenAI says GPT-6 can combine text, visuals and interactive parts such as buttons, forms and charts. The company said rollout began October 7 for Plus, Pro, Business and Enterprise plans, with Free and Go starting October 8. Enterprise availability can depend on admin settings.
Haiku 5.5FAST, ROUTINE WORK
A smaller model with lower listed API prices
Anthropic positions Haiku 5.5 for high-volume tasks such as summaries, classification and database queries. Its published API prices for prompts up to 100,000 tokens are $0.10 per million input tokens and $0.50 per million output tokens. Anthropic estimates about 75% lower average cost than Haiku 4.5 across its request mix; actual task costs vary.
AI AssistantOPTIONAL WEB SEARCH
Acrobat can bring web sources into document questions
Adobe says Acrobat AI Assistant starts with your documents and lets you switch on web search when outside context would help. Web-sourced answers include clickable citations, and the search control can be turned off. We have not tested the feature. Adobe’s official document-productivity news page lists the October 7 announcement.
O3PAID DEVELOPER EDITS
Runway adds Kling video editing and Ideogram precision edits
Kling O3 edits existing clips up to 10 seconds; Ideogram 4.5 adds precise image edits. Runway Dev requests are paid, so check the per edit and per image costs first.
2.3.0WINDOWS · COMPATIBLE NVIDIA GPU
NVIDIA Broadcast 2.3 adds effects for recordings you already made
The File Processor adds supported prerecorded media to the app’s effects workflow. Requires Windows and compatible NVIDIA hardware; new upscaling and frame generation have higher GPU requirements.
Each update keeps its own check date and testing limits. Company claims are identified separately from our own results.
FEATURED UPDATE · GEMINI · OCTOBER 13, 2026
Google’s October 13 update expands access, but the Gemini app, AI Studio and API have different requirements.
Google says Gemini 4 Argon is now available to Google AI Pro and Ultra subscribers in the Gemini app and AI Mode in Search. Access differs across products: Google lists Ultra access for Antigravity and AI Studio, while developers use the paid Gemini API. The API pricing page excludes the Free tier and Google AI Pro plan.
Argon is now listed in the official model specifications. It accepts text, images, video, audio and PDFs and returns text; image generation and video generation are not supported. The sample video-inspection task described in our article is a proposed experiment, not a test we have run.
How the evidence changed
- September 25Our fact-check recorded what the public evidence showed before Google announced Argon. Read the October 13 Gemini 4 Argon access update and original evidence check.
- September 30Google announced Gemini 4 Argon and described a limited initial rollout through the Fairwind Program.
- October 6Our public API catalog check did not list Argon at that time.
- October 13Google announced broader app, Search, developer and enterprise availability. This is a later access update; it does not verify earlier anonymous tests.
We have not tested Argon’s accuracy or confirmed access in our account. The official pricing and restrictions explain which access routes require payment.
VISUAL EXPLAINER
What Google says Argon can do
These are explanatory graphics based on Google DeepMind’s capability descriptions. They are not Argon screenshots or generated outputs. We have not had hands-on access.
Find→02
Validate→03
Patch
BEYOND GOOGLE
More model news
Recent releases worth watching. These notes summarize the makers’ announcements; The Click Test has not run hands-on tests of these models.
Large 4PUBLIC PREVIEW · METERED API
Mistral Large 4 enters public preview
Mistral announced Large 4 for text, image, document, coding, and agent tasks. Its preview API is available in Mistral Studio; downloadable weights are planned for later in October. Mistral currently lists API prices of $0.68 per million input tokens and $2.09 per million output tokens. These are the maker’s specifications and prices, which may change. The Click Test has not run a hands-on test or used paid credits.
Read Mistral’s announcement and examples · Check model access and pricing
Sonnet 5.5EVERYDAY WORK · CODING · DOCUMENTS
Faster Sonnet for practical work
Anthropic says Sonnet 5.5 is more than 30% faster than Sonnet 5 and can cost up to 30% less per task. It emphasizes bug fixes and polished documents, slides, and spreadsheets. Anthropic lists access through Claude.ai on web and mobile, plus its developer platform.
SolCODING · COMPUTER USE · PROFESSIONAL WORK
A new Sol for complex tasks
OpenAI introduced GPT-6.1 Sol for coding, computer use, and professional work. It lists access for Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex, and through the OpenAI API. Availability may differ by product and plan.
Status checked October 7, 2026. Model specifications, speed, and cost claims above come from the makers, not independent The Click Test tests.
WHAT’S NEXT
What we’re watching
We’ll update this page when model makers publish meaningful releases or new access details. For claimed hands-on tests, look for the exact model name, a documented access route, and a repeatable result.
OUR STANDARD
How we handle updates
Every update is dated, linked to primary sources, and clear about what has and has not been verified. When evidence changes an earlier conclusion, we explain what changed. Read our editorial and testing policy.
More AI coverage
Browse AI & Codex Tools articles · Read the original Gemini 4 evidence check
Access and model catalogs can change. Each update keeps its announcement date and evidence status.