Baidu ERNIE

Foundation model family built by @Baidu_Inc.

🤗

[Baidu ERNIE @Baidu_ERNIE]

More to come!

[Tianxiang Sun @tianxiangsun]

Excited to introduce ERNIE 5.5 Flash Base, my first release since joining @Baidu_Inc !

An open-sourced 280B MoE model with just 14B active parameters, pretrained on 33.5T tokens. It delivers strong performance across reasoning, coding, and long-context benchmarks, with support for up to 1M context.

Huge respect and thanks to the incredible team behind this — outstanding work!

Post-training is already underway, and I can't wait to bring it to you soon.

And this is just the beginning — more exciting technical developments are on the way. Stay tuned!

https://t.co/7DBNYNwf81

[Baidu ERNIE @Baidu_ERNIE]

FP8 and BF16 weights are on Hugging Face under the ERNIE 5.5 Flash Base Model License Agreement v1.0.
Model: https://t.co/OuIfXW5sYy
Blog: https://t.co/2Bn4pgGFhP

[Baidu ERNIE @Baidu_ERNIE]

Introducing ERNIE-5.5-Flash-Base, the pre-trained base of ERNIE 5.5 Flash before any mid- or post-training. The post-trained ERNIE 5.5 Flash is coming soon.
• 280B total / 14B activated MoE
• 33.5T pre-training tokens
• 128K training context, 1M for fine-tuning and inference

[Baidu ERNIE @Baidu_ERNIE]

Evaluated on 24 short-context and 6 long-context benchmarks covering exam, knowledge, STEM, reasoning, multilingual, code and long context, against base models of similar size. Full results table on the model card.

[Baidu ERNIE @Baidu_ERNIE]

Pre-trained from scratch on 33.5T tokens in FP8, emphasizing repository-scale code and long context. Built for this run: a distributed Muon optimizer that scales to thousands of GPUs, and FlashMask-4, an attention kernel for large head dimensions.

[Baidu ERNIE @Baidu_ERNIE]

Baidu CoBuddy is live on @NovitaAI — a powerful coding model we helped bring to developers. Try it free 👇

[Novita AI @novita_labs]

🚀 Baidu CoBuddy is now available on Novita AI.

A code-focused model built for developers, coding agents, and complex software engineering workflows.

CoBuddy brings:

• Long-context coding support
131K context window and up to 65K output tokens

• Agent-ready capabilities
Native tool calling and reasoning support for AI agent workflows

• Fast, production-ready inference
High-throughput, low-latency access through Novita’s serverless API

Excited to bring Baidu CoBuddy to developers on Novita.

@ErnieforDevs @PaddlePaddle

[Baidu ERNIE @Baidu_ERNIE]

PaddleOCR-VL 1.6 is here, hitting a new SOTA at 96.33% on OmniDocBench v1.6.

Another big milestone for document intelligence 🎉

[PaddlePaddle @PaddlePaddle]

🚀PaddleOCR-VL 1.6 Officially Released!

We are thrilled to announce the official release of PaddleOCR-VL 1.6 — this version has set a new SOTA record of 96.33% on OmniDocBench, outperforming both open-source and proprietary solutions in text, formula, and table recognition.

📊 Key Highlights:
🔸 Ranked #1 on OmniDocBench v1.5 and Real5-OmniDocBench as well
🔸 Significant improvements in table, classic text, and rare character recognition
🔸 Enhanced seal, spotting, and chart recognition
🔸 Fully compatible with v1.5 architecture — zero migration, plug-and-play

From financial contracts and legal documents to research reports and historical archives — empower your document intelligence workflows by providing high-quality data to large language models (LLMs) and retrieval-augmented generation (RAG) systems.

✨ Industry-leading accuracy. Zero migration. Plug-and-play.

🔗 Read more: https://t.co/oNOfB6hbSY

#PaddlePaddle #PaddleOCR #AI #ComputerVision

[Baidu ERNIE @Baidu_ERNIE]

Really cool to see our models powering something so specialized and impactful.

[Baidu Inc. @Baidu_Inc]

Zack Williams brought one of the world's oldest writing systems to the ERNIE AI Developer Challenge: ancient cuneiform tablets.

Using PaddleOCR, he built NabuOCR to help read cuneiform from tablet images.

See the story behind @BoatbomberRBLX's winning project 👇

[Arena.ai @arena]

US vs China update. Stanford's AI Index put the US–China gap at 2.7%. Here's what two years of real-world use from the Text Arena shows.

Gap three years ago: +278. Today: +29.

@AnthropicAI's Claude Opus 4.6 Thinking vs. Baidu's
@ErnieforDevs Ernie 5.1 at the top.

The US has never lost #1, but the race keeps closing.

[Baidu ERNIE @Baidu_ERNIE]

We're also launching the ERNIE Ambassadors Program! 🌍 We're looking for developers, creators, researchers, and community builders worldwide to help explore and build with ERNIE models together.

📩 Apply here: https://t.co/aU3qUDGmZt

[Baidu ERNIE @Baidu_ERNIE]

ERNIE 5.1 is here 🚀

ERNIE 5.1 significantly reduces pretraining cost while compressing total parameters to ~1/3 and activated parameters to ~1/2 — using only ~6% of the pretraining cost compared to models at similar scale, while achieving leading performance in its class.

💡Key highlights:

1/ Strong agentic performance approaching leading frontier models. ERNIE 5.1 surpasses DeepSeek-V4-Pro on both τ3-bench and SpreadsheetBench-Verified.

2/ Strong world knowledge and creative writing capabilities, with GPQA and MMLU-Pro performance approaching leading closed-source models, and creative writing ability nearing Gemini 3.1 Pro.

3/ Frontier-level reasoning performance. ERNIE 5.1 scores 99.6 on the challenging AIME26 benchmark with tools, second only to Gemini 3.1 Pro.

4/ Deep search capability. On May 9, ERNIE 5.1 ranked #4 globally and #1 among Chinese models on the Arena Search leaderboard with a score of 1223.

ERNIE 5.1 is now available on ERNIE and the Baidu AI Studio Model Playground:
👉https://t.co/qhd67Lg3B4
👉https://t.co/AaQSqDmVGU
👉https://t.co/uCNiypIu1q

[Baidu ERNIE @Baidu_ERNIE]

@xerxes233 @arena The API is currently in closed beta. You can join our Discord and apply for whitelist access to test it for free!https://t.co/LREVPWs9k5

[Baidu ERNIE @Baidu_ERNIE]

Another update from @arena 👀

ERNIE 5.1 is now ranked #4 in Search Arena — making ERNIE one of the top-performing labs in Search and currently the only Chinese model in the Top 10.

Official release coming very soon 🚀

[Baidu ERNIE @Baidu_ERNIE]

@DwiAtmika7 Baidu has launched the Cobuddy model — it’s currently available for free on OpenRouter
https://t.co/StBQfkOa2A

[Baidu ERNIE @Baidu_ERNIE]

EERNIE 5.1 introduces several key technical innovations:

1/ Multi-Dimensional Elastic Pre-Training for dramatically improved parameter efficiency and lower training cost

2/ Decoupled Fully-Asynchronous RL Training for more scalable and stable long-horizon agentic RL

3/ Multi-stage RL pipeline centered on OPD for efficient capability fusion across domains

For full details on the architecture and training methodology, read the technical report:
https://t.co/21PfUvD9NI

[Baidu ERNIE @Baidu_ERNIE]

Stay tuned for some exciting updates coming your way soon 🚀

[Baidu Inc. @Baidu_Inc]

Baidu Create 2026 is coming up fast, and the agenda is packed!

Beyond the main forum, our flagship developer conference returns with special forums on AI infrastructure, agent development, real-world applications, and more — plus plenty to explore on site.

Take a look 👇

[Baidu ERNIE @Baidu_ERNIE]

@Chahatxsharma @arena More detailed model info and benchmarks will be shared very soon. Watch this space! 👀

[Baidu ERNIE @Baidu_ERNIE]

Introducing ERNIE 5.1 Preview — now live! 🚀
Ranked #13 globally and #1 among Chinese labs on @arena 's Text Arena.
Top-10 worldwide across:
📐 Math — #9
⚖️ Legal & Government — #1
💼 Business, Management & Financial Ops — #4
💻 Software & IT Services — #7

Built on the strong pre-training foundation of ERNIE 5.0, ERNIE 5.1 Preview compresses total parameters to ~1/3 and activated parameters to ~1/2 of its predecessor, while using only ~6% of the pre-training cost of comparable models — delivering leading foundational performance at its scale.

Try it now 👉 https://t.co/jNwEbwrKps
More new models are on the way — stay tuned 👀