OpenAI Assistants API migration deadline: August 26, 2026
Complete migration to the Responses API and Conversations API, including tool and thread-state tests, before shutdown.
Verified AI launch intelligence · Creator distribution
What launched. What changed. What it costs. Whether it worked.
- Launches.
- Changes.
- Pricing.
- Evidence.
- Distribution.
Kingy AI tracks AI product launches, model changes, pricing and tested workflows, then separates official facts, company claims and evidence from what remains unproven.
Home of a large AI-focused YouTube channel — 2M+ subscribers and 94.3M views.
@kingy-ai · Watch on YouTube
Right now
Trending on Kingy
Three stories readers are opening now.
Cost intelligence · Research beta
What AI models cost—verified at the source.
Compare normalized API rates and estimate a real workload with the existing Kingy calculator.
10 source-checked observations · 4 providers · 11 official sources · verified August 17, 2026
Operational intelligence
Open Stack Change Radar
Know what changes next in your AI stack.
Verified vendor deadlines, replacements and required actions—available as a page, RSS, JSON and calendar feed.
Google retires Gemini Robotics ER 1.6 Preview on August 31
Replace the 1.6 preview model ID with gemini-robotics-er-2-preview and validate robotics behavior, tool calls, media inputs and safety controls before August 31.
Google lists October 16 as Gemini 2.5 Pro’s earliest possible shutdown date
Move workloads to gemini-3.1-pro-preview and validate preview-model suitability, behavior and capacity before cutover.
How Kingy works
From launch claim to useful evidence.
Kingy AI records what changed, checks the source, tests workflows where practical, and connects credible products with creators and buyers.
What launched or changed
Product releases, model updates, funding, availability and material feature changes, linked to their sources.
Explore AI launchesWhat it costs and who confirmed it
Pricing, access and capability notes are attributed to official sources, company claims or named external evidence.
Browse product recordsWhether the workflow held up
Kingy-tested examples show the inputs, outputs and limits observed. Untested claims stay labelled as claims.
See workflow evidencePut credible products in context
Creator videos, launch records, product profiles and supporting assets help qualified AI products reach the right audience.
Request a distribution fit reviewBuilt for the AI market
Use Kingy AI to decide, prove or distribute.
Decide what is worth using
Follow model changes, compare current costs, inspect product records and use source-linked workflows instead of launch-day claims alone.
Track launches and changesBuild a product record and launch path
Submit a launch or material product change for editorial review, connect releases to evidence, and request creator distribution when the product is ready to demonstrate.
Explore the founder pathPlan with inspectable campaign evidence
Review public creator examples, measurement boundaries, ROI planning inputs and disclosure standards before choosing a campaign.
Review campaign evidenceEvidence status
Read the standards“Verified” tells you what the evidence supports.
A source can confirm that something launched. It does not automatically prove that every product claim worked in practice.
- Official sourceConfirmed by primary documentation or an announcement.
- Company claimAttributed to the vendor, without presenting it as independent proof.
- Kingy-testedA workflow or output reproduced and described by Kingy AI.
- Third-party evidenceSupported by a named external source readers can inspect.
- Not yet verifiedAn open claim, missing source or result that still needs evidence.
Research
Read the indexHow much traffic does AI actually send?
Published estimates disagree by more than an order of magnitude. Wave One compares nine study records selected under its published rules and adds Kingy's own measured seven-day evidence, with limits and corrections shown.
- 677.8 : 1Pages AI crawlers took from kingy.ai for every visit they sent back.
- 1.3279%Share of our own traffic that came from AI assistants, 20–26 July 2026, seven complete days.
- 9 studiesCatalogued by measurement method, tiered by whether the method can be checked.
- CC BY 4.0Full dataset downloadable as CSV; chart as SVG and PNG. Republish with attribution.
01 · Track
See all launches
Launch Radar
Aug 13
DeepSeek Harness developer preview DeepSeek opened the DeepSeek Harness developer preview and published its source code on August…
READ
Aug 7
gamedai live AI sports radio gamedai says its live AI sports-radio service went live on August 7, 2026 for…
READ
Aug 4
Liquid AI LFM2.5-2.6B Liquid AI released the base and post-trained LFM2.5-2.6B checkpoints on August 4, 2026. The…
READ
Jul 31
DeepSeek V4-Flash-0731 launches in API beta with native Codex support DeepSeek released DeepSeek-V4-Flash-0731 as the official V4-Flash API public beta. The checkpoint keeps the…
READ
Jul 30
Google retires Gemini Robotics ER 1.6 Preview on August 31 Google announced that gemini-robotics-er-1.6-preview will shut down on August 31, 2026, and its deprecation…
READ
Watch
Watch on YouTube
On YouTube now
AI product launches, changes, costs and documented workflow tests.
02 · Inspect
Browse product records
Featured product records
Breakthrough Watch
Explore the verification tracker
What’s verified—and what still isn’t—in mathematics and science.
85 source-linked records with explicit evidence tiers, review states, limitations and correction histories.
Independent preprints claim Pólya’s Neumann bound for every Euclidean ball
Two independently developed preprints claim Pólya’s lower bound for the Neumann eigenvalue-counting function of every Euclidean ball. Together with the published Dirichlet result and Neumann-disk case, the…
Review evidenceA formally verified proof package for the Cycle Double Cover Conjecture
The proof package establishes that every finite loopless bridgeless undirected multigraph has a multiset of cycles covering every edge exactly twice. A Lean formalization and two independent…
Review evidenceThe Jacobian conjecture is false in dimension 3
An explicit polynomial map from C³ to C³ has constant Jacobian determinant −2 but is not injective. The example disproves the unrestricted Jacobian conjecture in dimension 3.
Review evidencePlan an AI workflow with costs and evidence in view.
Answer five questions to map the categories your workflow needs, the tradeoffs to expect and a 30-day starting plan. Missing pricing or capability evidence is clearly flagged.
Five questions
Browser-based
ChatGPT Passport: Plans, Privacy & Data Boundaries
ChatGPT is OpenAI's general-purpose AI assistant for writing, research, coding, image creation and analysis, voice, and everyday work.
Free: Yes
API: Yes
Open: No
- Company
- OpenAI
- Pricing
- Free and paid plans are available; plan features, limits, and current prices are listed on ChatGPT's official pricing page.
Claude Passport: Plans, Privacy & Account Boundaries
Claude is Anthropic's AI assistant for writing, analysis, coding support, document work, and long-context reasoning for individuals, teams, and enterprises.
Free: Yes
API: Yes
Open: No
- Company
- Anthropic
- Pricing
- Claude offers free and paid plans; features, usage limits, and current prices vary by plan and region and are listed on Anthropic's official pricing page.
Codex
Codex is OpenAI's coding agent for writing, reviewing, and shipping code across local projects, editors, terminals, cloud tasks, and automated workflows.
Free: Yes
API: No
Open: No
- Company
- OpenAI
- Pricing
- Codex access and usage limits vary by ChatGPT plan and account; current eligibility and prices are listed on OpenAI's Codex and ChatGPT pricing pages.
Perplexity Passport: Plans & Data Boundaries
Perplexity launched Deep Research as a research mode that runs many searches, reads sources, reasons through the material, and returns a structured report.
Free: Yes
API: No
Open: No
- Company
- Perplexity AI
- Pricing
- At launch, Perplexity said Deep Research would be free with limited daily use for non-subscribers and higher usage for Pro subscribers.
Lovable
Lovable 2.0 added a refreshed product experience with workspaces, multiplayer collaboration, chat mode agent, security scan, dev mode, visual edits, custom domains, and pricing changes.
Free: Yes
API: No
Open: No
- Company
- Lovable
- Pricing
- Lovable 2.0 listed Pro plans starting at $25/month and Teams plans starting at $30/month for teams.
Runway
Runway’s practical boundary is the exact web-app plan: credit allowances and rollover differ, API credits remain separate, and commercial-use language does not replace input rights.
Free: Yes
API: Yes
Open: No
- Company
- Runway
- Pricing
- Runway states that web-app plans and credits are separate from API credits and cannot be transferred between those products.
Canva AI
Canva AI brings generative design, writing, image, presentation, and editing tools into Canva.
Free: Yes
API: No
Open: No
- Company
- Canva
- Pricing
- Canva offers free and paid plans; AI feature limits vary by plan.
03 · The report
View all AI News
Latest AI News
AI News
How to Choose an AI Inference Provider
A source-audited buyer's guide to choosing AI inference by accepted-job cost, quality, latency, capacity, privacy, reliability, operations, and exit risk.
AI News
AI Inference Providers Compared: Models, Pricing, Speed, Limits, Privacy, and Availability
An evidence-led comparison of 20 AI inference offerings across model access, pricing, measured speed, limits, privacy, residency, capacity, and deployment modes.
AI News
Qwen3.8-27B Beats GPT-5.6 Terra on the Agentic Index—By Less Than One Point
Qwen3.8-27B has edged past GPT-5.6 Terra and several larger models on the Artificial Analysis Agentic Index. Here is what the score proves, what it does not, and how the model fits on one 24GB RTX 3090 or 4090.
AI News
Did Cursor Just Kill GitHub? Inside Origin’s Bid to Rebuild Code Hosting for AI Agents
Cursor Origin does not kill GitHub. This evidence-led analysis separates confirmed beta features from open questions and gives practical advice for developers and engineering teams.
AI News
Grok Build vs. Codex vs. Claude Code: Which Is Best?
An evidence-led comparison of Grok Build, OpenAI Codex and Claude Code across coding benchmarks, specifications, cost, security and enterprise readiness.
AI News
How to Build a Private Local Coding Agent with Qwen3.8-27B
Build a source-audited Qwen3.8-27B local coding agent on one 24GB GPU with exact GGUF/runtime pins, strict Qwen Code permissions, privacy tests and an honest account of our inconclusive lab run.
Creator distribution
Show buyers what the product does—and what the evidence supports.
Review 132 public video records with source links and measurement boundaries.
Kingy turns launches, product changes and documented workflows into creator-led education. Paid relationships are disclosed, claims remain attributed, and campaign evidence is separated from editorial judgment.
7 campaign breakdowns published