OneKE
Open-source bilingual knowledge extraction engine via schema-guided LLMs
OneKE fills a real gap for bilingual knowledge extraction, especially for Chinese-English domains. The open-source model and toolchain are practical for research, but don't expect production-grade reliability without heavy fine-tuning. Compared to proprietary services like Google Cloud NLP or AWS Comprehend, OneKE offers full customization and transparency at the cost of lower out-of-the-box accuracy on narrow domains. Worth exploring if you need a customizable, research-oriented extraction pipeline.
- Researchers in information extraction and knowledge graphs
- NLP engineers building domain-specific knowledge bases in Chinese/English
- Academics needing a customizable, open-source extraction framework
- Developers prototyping bilingual NER/RE/EE systems
- Users requiring high accuracy on very specific narrow domains without fine-tuning
- Production systems needing real-time low-latency extraction
- Non-technical users expecting a no-code solution
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip OneKE if you need a production-ready, no-code knowledge extraction solution with guaranteed high accuracy and low latency, or if you cannot invest time in prompt engineering and fine-tuning.
Fine-tuning and prompt engineering require significant time and NLP expertise to achieve acceptable accuracy for narrow domains.
OneKE is entirely free and open-source, making it ideal for budget-constrained researchers and developers. Unlike proprietary services (e.g., Google Cloud NLP charges per request), OneKE has no per-call cost. However, you pay in setup time and compute resources. For teams needing quick, low-effort extraction, paid services may be more cost-effective despite per-call fees.
In short
OneKE — Open-source bilingual knowledge extraction engine via schema-guided LLMs. Best for Researchers in information extraction and knowledge graphs, NLP engineers building domain-specific knowledge bases in Chinese/English, Academics needing a customizable, open-source extraction framework. Free to use.
Viability Score
How likely is OneKE to still be operational in 12 months? Based on 4 signals — momentum (how recently it shipped), wrapper dependency, revenue model, and web presence.
Last calculated: July 2026
How we score →Key Features
- Schema-guided instruction-based knowledge extraction
- Bilingual support (Chinese and English)
- Multi-task extraction: NER, RE, EE
- Negative sample dictionary for accuracy improvement
- Polling instruction construction technique
- Integration with DeepKE-LLM and OpenSPG toolkits
- 4-bit quantization for low-power devices
- Unified JSON-like instruction format
- Support for schema descriptions and examples in prompts
- Open-source model and data on HuggingFace, ModelScope, WiseModel
- Trainable on ~0.4B tokens of instruction data
- Batchified instruction generation with dynamic schema count (4-6)
- Data deduplication and filtering heuristics
- Fine-tuned on Chinese-Alpaca-2-13B base model
About OneKE
OneKE is an open-source, bilingual (Chinese and English) knowledge extraction framework developed by Ant Group and Zhejiang University. It extracts structured knowledge from unstructured text using schema-guided instruction-based fine-tuning, supporting named entity recognition (NER), relation extraction (RE), and event extraction (EE). Built on Chinese-Alpaca-2-13B and trained on ~0.4B tokens from nearly 50 datasets, OneKE offers a unified JSON-like instruction format, negative sample dictionary, and batchified instruction generation. It integrates with DeepKE-LLM and OpenSPG toolkits for deployment and supports 4-bit quantization for low-power devices. While powerful for multi-domain, multi-task extraction, it requires careful prompt design and may produce inconsistent results or hallucinations. It's best suited for researchers and NLP engineers building domain-specific knowledge graphs, especially in bilingual contexts.
Behind the Verdict
OneKE is a strong choice for researchers and developers seeking an open-source, bilingual knowledge extraction framework. Its schema-guided approach allows flexible, multi-task extraction (NER, RE, EE) across domains. The inclusion of a negative sample dictionary and polling instruction strategy helps improve accuracy on semantically similar schemas. Integration with DeepKE-LLM and OpenSPG provides a complete toolchain, and 4-bit quantization enables deployment on low-power devices. However, the model's output is prompt-dependent; inconsistent results and hallucinations are possible. It struggles with very long documents due to context length limits, and production-level accuracy on narrow domains typically requires additional fine-tuning. The lack of a no-code interface and real-time low-latency support limits its appeal to non-technical users. Overall, OneKE is a valuable research tool but not a turnkey production solution.
Researching OneKE? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas OneKE actually fits — and what changes day-one when you adopt it.
You have a collection of 10,000 biomedical abstracts in English and Chinese. You define a schema with entity types (e.g., gene, disease) and relations (e.g., treats, causes). Using OneKE via DeepKE-LLM, you run batch extraction to populate the knowledge graph.
Outcome: Within a day of setup, you extract structured triples, yielding a preliminary knowledge graph that can be refined and integrated with OpenSPG.
You need to extract event triggers and arguments (e.g., 'acquisition' event: buyer, seller, date) from Chinese and English news. You create a schema and use OneKE's polling instruction method to handle dynamic event types.
Outcome: After a few hours of prompt tuning, you achieve reasonable extraction accuracy on a sample set, demonstrating feasibility before production deployment.
Use Cases
- Extract named entities from financial news articles in Chinese and English
- Build a knowledge graph from scientific literature using relation extraction
- Extract event triggers and arguments from news corpora for event understanding
- Create a domain-specific IE pipeline for medical records with custom schemas
- Generate structured data from bilingual legal documents for compliance analysis
Models Under the Hood
as of 2026-07-16
Limitations
- OneKE relies on schema-guided prompts, which can lead to inconsistent results if prompts are not carefully designed.
- The model is limited in context length, making it less effective for very long documents.
- It may also produce hallucinations or miss some entities due to model size constraints.
- The unified schema instruction structure may not cover all knowledge representation forms, leading to incomplete or inaccurate extraction in industrial applications with high coverage and accuracy requirements.
as of 2026-07-06
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published OneKE tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Open Source
$0
Ideal for
Researchers, academics, and developers who need a free, customizable bilingual knowledge extraction framework for experimentation and prototyping.
What this tier adds
The single open-source tier includes full model weights, inference code, and integration with DeepKE-LLM and OpenSPG toolkits, with no usage limits.
Where the pricing makes sense
The company stage and team size where OneKE's pricing actually pencils out — and where peers do it cheaper.
OneKE is entirely free and open-source, making it ideal for budget-constrained researchers and developers. Unlike proprietary services (e.g., Google Cloud NLP charges per request), OneKE has no per-call cost. However, you pay in setup time and compute resources. For teams needing quick, low-effort extraction, paid services may be more cost-effective despite per-call fees.
Setup time & first value
How long it actually takes to get something useful out of OneKE — broken out by persona, not the marketing-page minute.
For a researcher familiar with Python and HuggingFace, you can download the model and run your first extraction in under an hour. Integrating with DeepKE-LLM or OpenSPG may take a few more hours of environment setup. Non-technical users may require 1-2 days to understand the instruction format and prompt engineering.
Switching to or from OneKE
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From rule-based IE systems: OneKE's schema-guided approach can replace complex rule sets with flexible prompts, but you'll need to convert your schema definitions to the JSON-like format.
Integrations
Resources & Guides
Official links
Tools that pair well with OneKE
Common stack mates teams adopt alongside OneKE, with the specific reason each pairing earns its keep.
Featured Head-to-Head Comparisons
Alternatives to OneKE
View all
SheetAI.app
Run AI text generation, classification, and extraction inside Google Sheets with simple formulas.
OpenAgents
Open-source platform for deploying language agents in everyday scenarios.
Frequently Asked Questions
Categories
Best-of guides
Used OneKE? Help shape our editorial sentiment research.