What Happened to PromptPerfect? Jina AI Acquisition by Elastic Explained

If you used PromptPerfect and woke up to a shutdown banner, the story probably sounds random. A prompt optimizer gets acquired by a search company and dies nine months later. What?

It isn't random. Elastic didn't buy PromptPerfect. They bought Jina AI's embedding models, rerankers, and research team — then decided the consumer prompt app wasn't worth maintaining. This post walks through the full chain: who Jina was, what Elastic wanted, why PromptPerfect lost, and what happens next.

The short version

October 9, 2025: Elastic completes acquisition of Jina AI.

March 2026: Jina announces PromptPerfect shuts down September 1, 2026.

June 2026: New PromptPerfect signups close.

September 1, 2026: PromptPerfect goes offline.

October 1, 2026: User data deleted.

Jina's search products (Embeddings, Reranker, Reader) continue under Elastic. PromptPerfect does not.

Who was Jina AI?

Jina AI was a Berlin-based search AI company founded in 2020 by Han Xiao. Before Jina, Han worked on search infrastructure at Tencent's AI Lab (WeChat search) and created open-source tools like bert-as-service and Fashion-MNIST.

Jina started as a neural search company. In April 2020 they open-sourced Jina, a framework for building multimodal search pipelines — image-to-image, text-to-text, code-to-code. The pitch was "Google for your data": search anything, not just web pages.

Over five years the product line shifted toward foundation models for retrieval:

• Jina Embeddings (v2, v3, later v5 omni) — dense vector models for semantic search and RAG
• Jina Reranker — reranking retrieved documents for better relevance
• Jina Reader — converts any URL to LLM-friendly text via r.jina.ai (free API, widely used in RAG pipelines)
• Small language models for grounding tasks like HTML-to-Markdown conversion

PromptPerfect was a side product: a web app that optimized prompts for ChatGPT, Claude, Midjourney, and Stable Diffusion. It launched around 2023 and built a loyal user base. But it was never the technology Elastic paid for.

What Elastic actually bought

Elastic (NYSE: ESTC) runs Elasticsearch — one of the most widely deployed search engines in enterprise IT. Since 2023 they've pushed hard into "Search AI": vector search, RAG, and context engineering for LLM applications.

Jina AI fit that roadmap precisely. Elastic's October 9, 2025 announcement lists what they acquired:

• Multimodal and multilingual embedding models (text and images, single-vector and multi-vector outputs)
• Advanced rerankers for visual documents, long multilingual text, and code search
• Small language models for HTML-to-Markdown and HTML-to-JSON extraction
• Jina AI's research team, including Han Xiao as VP of AI at Elastic

Elastic CEO Shay Banutzi framed it simply: "Search is the foundation of generative AI." The acquisition adds Jina's models alongside Elastic's existing ELSER sparse embedding model and vector database inside Elasticsearch.

Models stay on Hugging Face for free download. Enterprise customers get them through Elastic Inference Service (EIS) on Elastic Cloud — run embeddings and rerankers natively next to Elasticsearch without managing GPU infrastructure.

None of this requires a consumer prompt optimizer.

Why PromptPerfect doesn't fit Elastic's plan

PromptPerfect solved a different problem for a different buyer.

Elastic sells to enterprises: DevOps teams, security operations, observability, e-commerce search, RAG pipelines inside Fortune 500 apps. Their customers pay for relevance, scale, and SLAs — not for polishing Midjourney prompts.

PromptPerfect sold to individuals and small teams: marketers, creators, developers who wanted better ChatGPT outputs. B2C product, B2C pricing ($0–20/month for most users), separate codebase from the embedding stack.

After acquisition, every engineering hour on PromptPerfect is an hour not spent integrating jina-embeddings-v3 into Elasticsearch 9.x or shipping v5 omni models on EIS. Elastic had no incentive to keep it.

Common outcomes for acquired consumer products: spin-off, open-source, or sunset. Jina chose sunset. No public announcement of a buyer, no open-source handoff, no "PromptPerfect will live on as an independent company." Just a shutdown notice on promptperfect.jina.ai.

That's frustrating if you paid for Pro. It's also predictable if you understand what Elastic's shareholders optimize for.

Timeline: from startup to shutdown

2020 — Han Xiao leaves Tencent, founds Jina AI in Berlin. Open-sources Jina neural search framework.

2022–2023 — Jina pivots toward embedding models and search APIs. PromptPerfect launches as a prompt optimization tool for LLMs and image generators.

September 2024 — jina-embeddings-v3 released. 570M parameters, multilingual, 8192 token context, competitive with OpenAI and Cohere on MTEB benchmarks.

2024–2025 — Jina Reader (r.jina.ai) becomes widely used in RAG pipelines. Major LLM companies reportedly use it for web crawling and training data prep.

October 9, 2025 — Elastic completes acquisition of Jina AI. Press release focuses entirely on embeddings, rerankers, and search infrastructure. PromptPerfect not mentioned.

March 14, 2026 — Shutdown notice appears on promptperfect.jina.ai.

Early June 2026 — New PromptPerfect signups disabled.

September 1, 2026 — PromptPerfect goes offline. API, Arena, Chrome extension, saved libraries: all gone.

October 1, 2026 — User data deleted (~30 days after shutdown).

What happens to other Jina products

If you used Jina for search, not prompts, your stack mostly continues.

Jina Embeddings and Reranker: available on Hugging Face. Enterprise deployment via Elastic Inference Service. jina-embeddings-v3 and v5 omni models integrate into Elasticsearch semantic search and RAG workflows.

Jina Reader (r.jina.ai): still active as of mid-2026. Free tier with rate limits. Used to fetch clean text from URLs for LLM input.

jina.ai website: redirects focus toward Elastic's search AI narrative. Company page states acquisition date: October 9, 2025.

Han Xiao: VP of AI at Elastic. Public writing shifts to search foundation models and retrieval, not prompt optimization.

PromptPerfect: dead. No migration path to another Jina product. Export your data and move to a third-party tool.

Did Elastic have to kill PromptPerfect?

Legally, no public requirement. Practically, yes — from a business standpoint.

Spinning off PromptPerfect would mean finding a buyer or founding team, negotiating terms, splitting infrastructure, and maintaining two brands. Elastic had no reason to bother. The product didn't drive Elasticsearch adoption.

Open-sourcing PromptPerfect would cost little but still needs maintenance, security patches, and support. Elastic's brand is enterprise search, not consumer AI tools.

Keeping it running inside Elastic means staffing a product line outside core strategy. Large acquirers rarely do this unless the product has meaningful revenue aligned with the parent company's buyers. PromptPerfect didn't.

Users feel this as neglect. Elastic's investors probably see it as focus.

What PromptPerfect users should do

Nothing changes the shutdown date. You can still influence how painful the transition is.

1. Export your prompt library before September 1 (Settings → Export Data, or manual copy). We have a step-by-step export guide on this blog.
2. Cancel billing so you don't pay for months after the product is dead.
3. Pick a replacement for Auto-Tune-style optimization. If you pasted rough prompts and got structured output back, promptmake.net/text covers the same loop for free — Generate from scratch or Enhance an existing draft, model-specific output for ChatGPT, Claude, Midjourney, FLUX, Sora, and more.
4. If you used Jina Reader or Embeddings in a RAG pipeline, check Elastic's docs for EIS migration. That's a separate workflow from PromptPerfect.

How this compares to other AI acquisitions

Acquirers usually keep what aligns with their core business and drop the rest.

When larger companies buy AI startups, the pattern repeats: buy the model team and infrastructure, sunset the consumer app. Users of the consumer product get 6–12 months notice if they're lucky.

PromptPerfect users got roughly six months from public announcement to shutdown. Better than an overnight 404. Worse than a spin-off with continued service.

The lesson for anyone building workflows on a free or cheap AI tool: if the parent company gets acquired by an enterprise vendor with a different customer profile, assume your product is at risk. Export regularly. Don't treat a SaaS prompt library as permanent storage.

Frequently asked questions

When did Elastic acquire Jina AI?

October 9, 2025. Elastic issued a press release and blog post confirming the completed acquisition that day.

Why did Elastic buy Jina AI?

For embedding models, rerankers, and search foundation model research. Elastic integrates these into Elasticsearch for vector search, RAG, and context engineering in enterprise applications.

Is Jina AI shutting down entirely?

No. Jina's search products continue under Elastic. The Jina AI brand persists on jina.ai. Only PromptPerfect (the consumer prompt optimizer) is being sunset.

Why is PromptPerfect shutting down?

It was a B2C prompt tool outside Elastic's enterprise search focus. After acquiring Jina for embeddings and rerankers, Elastic chose not to maintain PromptPerfect.

Who founded Jina AI?

Han Xiao, in 2020. Previously at Tencent AI Lab. Creator of bert-as-service and Fashion-MNIST. Now VP of AI at Elastic.

Will PromptPerfect be open-sourced?

Nothing public suggests that as of June 2026. Plan as if the product is gone permanently after September 1, 2026.

What replaces PromptPerfect for prompt optimization?

For manual optimization, tools like PromptMake (/text for typed prompts, /image for photo-to-prompt) cover similar workflows without subscription. See our comparison article on this blog for a full breakdown.

Does the acquisition affect Jina Reader (r.jina.ai)?

Not directly related to PromptPerfect's shutdown. Jina Reader remains a separate product focused on URL-to-text for RAG. Check jina.ai for current status and rate limits.

Bottom line

PromptPerfect didn't fail because prompt optimization stopped mattering. It failed because Elastic bought the search infrastructure company behind it and had no use for a consumer prompt app.

Jina AI continues as part of Elastic's Search AI platform. PromptPerfect does not. Export your prompts, find a replacement before September, and treat this as a reminder that free tools attached to venture-backed startups don't last forever — especially after acquisition by an enterprise vendor with different priorities.