Release history
Changelog
On this page
The short answer
- This page lists public Litmiq site and app releases, with dates taken from QA records and release notes.
- It does not invent traffic, ratings or unpublished work.
- A machine-readable feed is at /changelog.xml.
Releases
v7.3: Templates library, glossary pages, integrations, dataset and changelog
This release adds a prompt-eval templates library with one runnable suite per use case, integration guides for testing prompts from LangChain, LlamaIndex, the Vercel AI SDK and GitHub Actions (no official vendor integrations are claimed), per-term glossary pages, topic hubs, a versioned benchmark dataset landing page, additional honest Litmiq vs pages, a public changelog, branded share report views, embed snippets for the calculator and shutdown tracker, and a template for a monthly reliability note. One canonical product sentence is now used in the site, schema,
llms.txtand the listings copy file.v7.1: Run-this-example buttons, key numbers, model-vs-model pages
Guides and reports gained a one-click “Run this example in Litmiq” bar, a short dated “Key numbers” list sourced from Litmiq’s own CC BY data or official provider pages, and a visible “Updated” date. Fifteen head-to-head pages under the benchmark use only its measured per-case results. Production id
87b71d04, released 17 November 2026.v7.0: In-browser demo, four free tools, deprecation alerts
A 30-second demo on the homepage and in the app that replays a real recorded run (three open models on Cloudflare Workers AI, 13 November 2026; labelled as a recording, not live), a free tools hub (token and cost calculator, prompt diff checker, JSON schema validator, model deprecation tracker), and an optional model-shutdown alert signup. Production id
be4ceac3, released 17 November 2026.v6.9.4: Week 5 guide: testing summarization prompts
Published Testing summarization prompts: 8 illustrative cases for changed numbers, dropped hedges, lost negations, invented owners and instructions hidden in the source, with downloadable CSV/JSON. Production id
aa5f251a, released 17 November 2026.v6.9.3: Light-mode homepage fixes
The homepage product mock in light mode uses light surfaces instead of a hardcoded dark inset. Light syntax colours meet WCAG AA on near-white code panels. Container gutters were restored so section headings are not flush to the viewport. The footer “Open the app” button uses brand purple in light mode, not near-black. Production id
cd11038a, fromqa/v6.9.3/RELEASE-v6.9.3.md, released 13 November 2026, 4:57 PM America/New_York.v6.9.2: SEO audit fixes
Meta description lengths, JSON-LD Person and SoftwareApplication, font preload, a crawlable lead on
/app, hash-based lastmod, an RSS feed, and homepage guide links. Staging QA record dated 13 November 2026.v6.9.1: Shared footer on every marketing page
One redesigned footer replaced four legacy variants on every page except the app UI (the 404 page gained the footer too). Production id
4d44b558, fromqa/v6.9.1/RELEASE-v6.9.1.md.v6.9: Suite links, new comparisons, CI guide, Anthropic in the app
The app gained
/app?suite=and#suite=loaders, Import/Share/Export of YAML, and shareable result links packed into the URL fragment. New pages included Run in Litmiq, Promptfoo without a CLI, Litmiq vs Braintrust, DeepEval and Langfuse, prompt testing in CI/CD, and Compare OpenAI models. The app’s model list expanded to include Claude, with keys still held in the browser. Production deployment iddfdb55e4, recorded inqa/v6.9/RELEASE-v6.9.md.v6.8: OpenAI Evals migration and model-migration testing
Added a guide and in-browser converter for the OpenAI Evals shutdown (read-only since 31 October 2026, scheduled to shut on 30 November 2026), a model-migration testing guide, and an upgraded LLM-as-a-judge guide. Staging QA record dated 13 November 2026.
v6.5–v6.7: Original research: reliability benchmark and State of 2026 reports
v6.5 published the LLM Reliability Benchmark 2026 (12 models, 40 deterministic tests on Cloudflare Workers AI, CC BY 4.0 dataset). v6.6 added the AI chatbot failures study. v6.7 added three State of 2026 reports — laws, API pricing, and customer support statistics — plus the reports hub. Dates are from staging QA records performed on 12 November 2026, matching the benchmark’s measurement date.
v6–v6.4: SEO and GEO pages, brand, pricing, Ask Litmiq, light mode
v6 added comparison and glossary pages, an entity graph, Open Graph images and
llms-full.txt. v6.1 was a brand refresh. v6.2 added the pricing page and Pro waitlist, a guides redesign and a legal pass. v6.3 added the Ask Litmiq chatbot. v6.4 added site-wide light mode and a persistent theme toggle. Dates are from staging QA records performed on 11 November 2026.v2–v5: Cloudflare Pages candidate, SEO articles and templates
The site moved onto a Cloudflare Pages candidate (static, analytics off) with form handling via Web3Forms. v3 polished the web and mobile UI. v4 and v4.1 added the first published SEO articles and a font layout fix. v5 introduced the current guide layout, loadable test-case templates, and
llms.txt. Dates are taken from staging QA records performed on 10 November 2026.
Feeds
Subscribe to changelog.xml for releases, or to feed.xml for guides and reports (which also includes recent changelog entries).
FAQ
Is this every commit?
No. It is the public release history taken from QA release notes, staging QA records and the build modules, from the Cloudflare Pages candidate onwards.
Why are several versions on the same day?
The v6.8–v6.9.3 and later work was shipped on 13 November 2026. Each version is listed separately so the history stays factual.
Where is the RSS?
Release items are in /changelog.xml and are also included in the site feed at /feed.xml.
See what you can run today
Templates, tools and the app are linked from the footer on every marketing page.
- Free, no signup
- Runs in your browser
- Your OpenAI key, sent only to OpenAI