Baichuan Intelligence
Baichuan Intelligence is a Beijing-based AI foundation model company founded in 2023 that develops proprietary general-purpose and medical-specialized large language models. It serves Chinese enterprises via usage-based APIs and healthcare institutions via its medical LLM family and free clinical-decision-support program.
- Company typePrivate
- Founded2023
- HeadquartersHaidian, China
- Headcount101–250
- GTM typeB2B and B2C
- OfferingSoftware
What Baichuan Intelligence does
Baichuan Intelligence (百川智能) is a Beijing-based AI foundation model company founded on March 24, 2023, by former Sogou CEO Wang Xiaochuan. The company develops large language models from scratch across two parallel tracks: a general-purpose enterprise line (Baichuan4, Baichuan4-Turbo, Baichuan4-Air) and a vertically specialized medical line (Baichuan-M1, M2, M3, M4). Its technical stack is built on Transformer and proprietary Mixture-of-Experts (PRI) architectures, with proprietary training innovations including evidence anchoring (95% precision match), large verifier systems, and patient simulation for clinical reasoning. Domain-specific training corpora reach 20 trillion tokens for the medical family and 2.6 trillion multilingual tokens for general models.
The company's business model combines open-source community-driven adoption with commercial monetization. Foundation models (Baichuan-7B/13B, Baichuan2-7B/13B) are released under free commercial licenses for organizations below 1 million DAU, generating downloads and ecosystem momentum on HuggingFace and GitHub. Enterprise revenue is generated through usage-based API pricing — Baichuan4-Air at 0.98 fen per 1,000 tokens and Baichuan4-Turbo at 80% of GPT-4o pricing. The company also operates a consumer healthcare app, Bai Xiao Yi (百小医), an AI family doctor for symptom triage and medical record management, and runs the 'Hail to All Seas' (海纳百川) free medical API program targeting Chinese healthcare institutions. Revenue is geographically restricted to mainland China.
The company is recognized as one of China's 'Six Little Tigers' AI unicorns and has raised approximately $740 million across three funding rounds, with strategic backing from Tencent, Alibaba, Xiaomi, and Ant Group, alongside government-backed AI industry investment funds from Beijing, Shanghai, and Shenzhen. HealthBench benchmark claims of surpassing GPT-5.2 with the Baichuan-M3 and M4 models represent the company's primary competitive positioning.
Baichuan Intelligence firmographics
Firmographics- Name
- Baichuan Intelligence
- Legal name
- 北京百川智能科技有限公司 (Beijing Baichuan Intelligence Technology Co., Ltd.)
- Website
- https://baichuan-ai.com
- Company type
- Private
- Founded year
- 2023
- Operating status
- Operating
- Headcount range
- 101–250 employees
- Short description
- Baichuan Intelligence is a Beijing-based AI foundation model company founded in 2023 that develops proprietary general-purpose and medical-specialized large language models. It serves Chinese enterprises via usage-based APIs and healthcare institutions via its medical LLM family and free clinical-decision-support program.
- Ownership category
- akta.pro rank
Baichuan Intelligence industry classification
Industry- Product category
- Large Language Models (AI Foundation Models)
- NAICS
- Professional, Scientific, and Technical Services (54)
- akta.pro primary industry
- Fine-Tuning, Adaptation & Custom Model Training (PEFT/LoRA/RLHF) (HDAAACAC)
- akta.pro secondary industry
- Speech Recognition, Dictation & Medical Transcription Platforms (HLACAAAI)
Keywords
Where Baichuan Intelligence is headquartered
LocationHeadquarters
- HQ city
- Haidian
- HQ country
- China
- HQ region
- Asia
Offices1 record
Markets served
Baichuan Intelligence business model
Business model- GTM type
- B2B and B2C
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations
Revenue model
- API Access & Enterprise Model Licensing: Baichuan Intelligence monetizes its LLM capabilities through commercial API access and enterprise model licensing. Enterprise-grade models (Baichuan4-Turbo, Baichuan4-Air) are offered at competitive price points — Baichuan4-Air at 0.98厘/千token, and Baichuan4-Turbo priced at 80% of GPT-4o's rate. The company also offers commercial licensing for its open-source models (Baichuan2-7B, Baichuan2-13B, Baichuan-7B, Baichuan-13B) with eligibility conditions (DAU below 1 million, non-SaaS/cloud provider applicants).
- Open-Source Community & Ecosystem: Open-source model releases on HuggingFace and GitHub build developer adoption and ecosystem momentum, potentially driving indirect monetization through brand influence, premium support services, and future enterprise conversions.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Usage-based | Pay-as-you-go | Baichuan4-Air — Enterprise-optimized MoE model at ultra-low cost |
| Usage-based | Pay-as-you-go | Baichuan4-Turbo — Enterprise high-availability model at competitive pricing |
| Freemium | Pay-as-you-go | Open-source models (Baichuan2-7B, Baichuan2-13B, Baichuan-7B, Baichuan-13B) — Free for eligible commercial use |
| Freemium | Pay-as-you-go | Baichuan-M3 Plus API — Free for medical institutions under 海纳百川 program |
Go-to-market motion2 records
Distribution channels4 records
Marketing channels4 records
Baichuan Intelligence product offering
Product offeringCore offering
Baichuan Intelligence develops and provides large language models (LLMs) for general-purpose and medical-specialized applications. The company offers enterprise-grade proprietary models (Baichuan4, Baichuan4-Turbo, Baichuan4-Air) via paid API access, open-source foundation models (Baichuan-7B/13B, Baichuan2-7B/13B) with free commercial licensing for eligible organizations, medical-specialized models (Baichuan-M1, M2, M3, M4) for clinical consultation, and a consumer-facing AI family doctor mobile application (百小医/Bai Xiao Yi). Revenue is generated through usage-based API pricing, commercial licensing, and free medical API access to qualifying healthcare institutions.
Product overview
Baichuan Intelligence offers a comprehensive portfolio of large language models spanning general-purpose and medical-specialized applications. The company provides a layered product architecture: core general models (Baichuan4, Baichuan4-Turbo, Baichuan4-Air) for enterprise use; specialized medical models (Baichuan-M3, Baichuan-M2, Baichuan-M1) designed for clinical consultation with evidence-based capabilities; and open-source foundation models (Baichuan-7B, Baichuan-13B, Baichuan2-7B, Baichuan2-13B) for developers. The company also offers consumer-facing products including 百小医 (Bai Xiao Yi) AI family doctor assistant and family health management bot. Founded in March 2023 by former Sogou CEO Wang Xiaochuan, the company has rapidly released over 10 models with millions of downloads.
Differentiator
Problem solved
Functional benefit
Brands
- 百小医 (Bai Xiao Yi): AI family doctor application that helps users prepare before medical consultations, analyze conditions after visits, interpret medical advice, and manage family health comprehensively.
- 百小应 (Bai Xiao Ying)
- Baichuan-M3
- Baichuan-M4
Products and services
- Baichuan4 General-purpose large language model ranked #1 domestically on SuperCLUE benchmark, featuring Search Agent capability, long context window, and industry-leading multimodal capabilities. Designed for enterprise knowledge Q&A, long text processing, and generative tasks, surpassing international mainstream models on Chinese tasks.
- Baichuan4-Turbo Enterprise-optimized version of Baichuan4 with 10%+ usability improvement, priced at 80% of GPT-4o with faster token response and flow speeds. Targeted at high-frequency enterprise scenarios.
- Baichuan4-Air MoE (Mixture of Experts) model using Baichuan's proprietary PRI architecture, designed for enterprise scenarios. Offers extreme cost efficiency at 0.98 cents per 1K tokens with performance exceeding international mainstream MoE models.
- Baichuan-M3 Open-source medical large model with 235 billion parameters, achieving #1 on HealthBench with 65.1 score and surpassing GPT-5.2 in medical consultation capabilities. Features strong reasoning and low hallucination for serious medical consultations.
- Baichuan-M2 Open-source medical LLM with 32B parameters using MoE architecture, deployable on a single 4090 GPU. Features 32K long context window and 60.1 HealthBench score, optimized for inference with better adaptation to Chinese medical environments than OpenAI open-source models.
- Baichuan-M1 First open-source medical enhanced model with 14 billion parameters, trained on 20T medical enhancement corpus with 32K context window. Deployable on single 4090 GPU.
- Baichuan2-13B Open-source commercially available multilingual model with 13 billion parameters, trained on 2.6T multilingual corpus with 4K context window. Shows 49% math, 46% code, 37% safety, and 25% logical reasoning improvements vs previous 13B model.
- Baichuan2-7B Open-source commercially available model with 7 billion parameters, achieving performance parity with 130B parameter LlaMA2 on English tasks despite 7B parameters. Trained on 2.6T multilingual corpus with 4K context window.
- Baichuan-13B Open-source bilingual Chinese-English model with 13 billion parameters, trained on 1.4T multilingual corpus with 4K context window.
- Baichuan-7B China's first open-source commercially available model with 7 billion parameters, trained on 1.2T multilingual corpus with 4K context window. Released within 100 days of company founding and achieved over 1 million downloads.
- 百小医 (Bai Xiao Yi) AI Family Doctor Consumer-facing AI family doctor mobile application for iOS/Android that helps users prepare for medical visits by organizing symptoms, analyze conditions post-visit, interpret medical advice, and manage family health with medication reminders and health tracking. Includes a family health management bot.
Quantifiable outcome
- Baichuan-M4 achieves HealthBench global #1 ranking, surpassing GPT-5.2 in medical capability
- +7 more outcomes
Companies that use Baichuan Intelligence
Customer profileNamed customers1 record
Segments1 record
Ideal customer profiles3 records
Baichuan Intelligence technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
AI capability5 records
Feature8 records
Baichuan Intelligence partnerships and signals
Strategic signalScale indicators4 records
Recent moves6 records
Expansion highlights6 records
Baichuan Intelligence competitors and assessment
Company assessmentBroad incumbents
- Anthropic: Global frontier-LLM incumbent behind Claude. Comparable to Baichuan in targeting enterprise and developer customers with safety- and accuracy-focused LLMs, but operating at global scale from a US base.
- OpenAI: Global frontier-LLM incumbent. Baichuan explicitly benchmarks against GPT-4o and GPT-5.2; OpenAI sets the global pricing and capability bar that Baichuan's cost-optimized models (Baichuan4-Air, Baichuan4-Turbo at 80% of GPT-4o) are positioned against.
- Tencent (Hunyuan): Tencent's Hunyuan model family is offered through Tencent Cloud. As a Baichuan investor and direct model competitor, Tencent represents both a strategic ally and an incumbent offering parallel APIs to overlapping Chinese enterprise customers.
- iFlytek: Established Chinese AI incumbent with deep roots in speech recognition and a broad portfolio spanning education, healthcare, and government. Competes with Baichuan in medical and enterprise AI but as a diversified incumbent, not a frontier-LLM specialist.
- Alibaba (Tongyi Qianwen): Alibaba's Qwen LLM family is offered through Alibaba Cloud alongside Baichuan's enterprise APIs. Although Alibaba is also an investor, its open-source Qwen models directly overlap with Baichuan's general-purpose and enterprise offerings.
Direct peers
- 01.AI: Chinese AI startup founded by Kai-Fu Lee, building Yi-series open and proprietary LLMs. Direct peer in the 'Six Tigers' cohort competing for Chinese enterprise and developer mindshare.
- Moonshot AI: Chinese AI startup behind the Kimi LLM, focused on long-context and consumer chat. Direct competitor in the 'Six Tigers' group, targeting overlapping Chinese enterprise and developer customers with comparable funding scale.
- Zhipu AI: Chinese AI startup developing the GLM/ChatGLM model family. Another 'Six Tigers' direct peer with overlapping enterprise GTM, comparable model cadence, and shared strategic investors (Tencent, Alibaba).
- DeepSeek: Chinese AI startup developing frontier open-source LLMs (DeepSeek-V3, R1). DeepSeek is one of Baichuan's closest direct peers in the 'Six Tigers' cohort, competing head-to-head on general LLM benchmarks and open-source developer adoption.
- StepFun (阶跃星辰): Chinese AI startup developing Step-series foundation models, frequently grouped with the leading wave of Chinese LLM companies. Direct competitor for enterprise API contracts and Chinese-language model leadership.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat6 records
Key risks6 records
Key highlights7 records
Customer concentration
Baichuan Intelligence social profiles
Digital presenceBaichuan Intelligence financial estimates
Financial estimateRevenue estimate
Valuation estimate
Baichuan Intelligence leadership team
Management profileNumber of profiles
Profiles1 record
Baichuan Intelligence funding detail
Funding detailFunding overview
Funding rounds3 records
Investors8 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Baichuan Intelligence M&A and investment
M&A and investmentM&A
Investments3 records
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Baichuan Intelligence
What does Baichuan Intelligence do?
Baichuan Intelligence develops and provides large language models (LLMs) for general-purpose and medical-specialized applications. The company offers enterprise-grade proprietary models (Baichuan4, Baichuan4-Turbo, Baichuan4-Air) via paid API access, open-source foundation models (Baichuan-7B/13B, Baichuan2-7B/13B) with free commercial licensing for eligible organizations, medical-specialized models (Baichuan-M1, M2, M3, M4) for clinical consultation, and a consumer-facing AI family doctor mobile application (百小医/Bai Xiao Yi). Revenue is generated through usage-based API pricing, commercial licensing, and free medical API access to qualifying healthcare institutions.
Is Baichuan Intelligence a public or private company?
Baichuan Intelligence is a private company. It is classified as venture growth investor backed and is currently operating.
When was Baichuan Intelligence founded?
Baichuan Intelligence was founded in 2023. It employs 101 to 250 people.
Where is Baichuan Intelligence based?
Baichuan Intelligence is headquartered in Haidian, China, in the Asia region.
How does Baichuan Intelligence make money?
Two revenue lines are on record. API Access & Enterprise Model Licensing is the primary driver. The others are open-Source Community & Ecosystem.
Who are Baichuan Intelligence's main competitors?
Broad incumbents on record are Anthropic, OpenAI, Tencent (Hunyuan), iFlytek and Alibaba (Tongyi Qianwen). Direct peers are 01.AI, Moonshot AI, Zhipu AI, DeepSeek and StepFun (阶跃星辰).
Does Baichuan Intelligence have an API?
Yes. Baichuan Intelligence offers API access for enterprise and developer use. The company provides Baichuan-M3 Plus API as part of their '海纳百川' (Hail to All Seas) program, offering free permanent access to the low-hallucination evidence-based medical large model API. Models are available through open-source platforms including Hugging Face (huggingface.co/baichuan-inc/Baichuan-M3-235B) and GitHub (github.com/baichuan-inc/). Commercial licensing is available through the business proposal page for enterprise use of open-source models including Baichuan2-7B, Baichuan2-13B, Baichuan-7B, and Baichuan-13B.
What industry is Baichuan Intelligence in?
Baichuan Intelligence's product category is Large Language Models (AI Foundation Models). Its primary akta.pro industry code is HDAAACAC, Fine-Tuning, Adaptation & Custom Model Training (PEFT/LoRA/RLHF), with a secondary code of HLACAAAI, Speech Recognition, Dictation & Medical Transcription Platforms. Its NAICS code is 54.