AI Research
Preferred on Google
Google News

Unified Embodied AI: Pelican-Unified 1.0

Pelican-Unified 1.0, the first unified embodied foundation model, achieves SOTA performance by integrating VLM, reasoning, and generation, proving unification enhances rather than compromises specialist strengths.

6 min read
Diagram illustrating the unified architecture of Pelican-Unified 1.0
Pelican-Unified 1.0's unified approach to embodied AI.

Visual TL;DR. Fragmented AI Models leads to Inefficiency & Limits. Inefficiency & Limits solves Pelican-Unified 1.0. Pelican-Unified 1.0 uses Unified VLM. Unified VLM enables Chain-of-Thought Reasoning. Chain-of-Thought Reasoning allows Simultaneous Optimization. Unified VLM leads to SOTA Performance. Chain-of-Thought Reasoning leads to SOTA Performance. SOTA Performance shows Unification Enhances.

  1. Fragmented AI Models: disparate specialized models for perception, reasoning, and action
  2. Inefficiency & Limits: leads to inefficiencies and limits holistic capabilities of AI systems
  3. Pelican-Unified 1.0: first unified embodied foundation model built on unification principle
  4. Unified VLM: single visual-language model maps diverse inputs to shared semantic space
  5. Chain-of-Thought Reasoning: autoregressive reasoning generates task- and action-oriented sequences in one pass
  6. Simultaneous Optimization: backpropagation of losses into shared representation enables simultaneous optimization
  7. SOTA Performance: achieves state-of-the-art performance by integrating perception, reasoning, generation
  8. Unification Enhances: proving unification enhances rather than compromises specialist strengths
Visual TL;DR
Visual TL;DR, startuphub.ai Pelican-Unified 1.0 uses Unified VLM. Unified VLM enables Chain-of-Thought Reasoning. Unified VLM leads to SOTA Performance. Chain-of-Thought Reasoning leads to SOTA Performance uses enables leads to leads to Fragmented AI Models Pelican-Unified 1.0 Unified VLM Chain-of-Thought Reasoning SOTA Performance From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Pelican-Unified 1.0 uses Unified VLM. Unified VLM enables Chain-of-Thought Reasoning. Unified VLM leads to SOTA Performance. Chain-of-Thought Reasoning leads to SOTA Performance uses enables leads to leads to Fragmented AIModels Pelican-Unified1.0 Unified VLM Chain-of-ThoughtReasoning SOTA Performance From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Pelican-Unified 1.0 uses Unified VLM. Unified VLM enables Chain-of-Thought Reasoning. Unified VLM leads to SOTA Performance. Chain-of-Thought Reasoning leads to SOTA Performance uses enables leads to leads to Fragmented AI Models disparate specialized models forperception, reasoning, and action Pelican-Unified 1.0 first unified embodied foundation modelbuilt on unification principle Unified VLM single visual-language model maps diverseinputs to shared semantic space Chain-of-Thought Reasoning autoregressive reasoning generates task-and action-oriented sequences in one pass SOTA Performance achieves state-of-the-art performance byintegrating perception, reasoning,generation From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Pelican-Unified 1.0 uses Unified VLM. Unified VLM enables Chain-of-Thought Reasoning. Unified VLM leads to SOTA Performance. Chain-of-Thought Reasoning leads to SOTA Performance uses enables leads to leads to Fragmented AIModels disparatespecialized modelsfor perception,… Pelican-Unified1.0 first unifiedembodied foundationmodel built on… Unified VLM singlevisual-languagemodel maps diverse… Chain-of-ThoughtReasoning autoregressivereasoning generatestask- and… SOTA Performance achievesstate-of-the-artperformance by… From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Fragmented AI Models leads to Inefficiency & Limits. Inefficiency & Limits solves Pelican-Unified 1.0. Pelican-Unified 1.0 uses Unified VLM. Unified VLM enables Chain-of-Thought Reasoning. Chain-of-Thought Reasoning allows Simultaneous Optimization. Unified VLM leads to SOTA Performance. Chain-of-Thought Reasoning leads to SOTA Performance. SOTA Performance shows Unification Enhances solves uses enables allows leads to leads to shows Fragmented AI Models disparate specialized models forperception, reasoning, and action Inefficiency & Limits leads to inefficiencies and limitsholistic capabilities of AI systems Pelican-Unified 1.0 first unified embodied foundation modelbuilt on unification principle Unified VLM single visual-language model maps diverseinputs to shared semantic space Chain-of-Thought Reasoning autoregressive reasoning generates task-and action-oriented sequences in one pass Simultaneous Optimization backpropagation of losses into sharedrepresentation enables simultaneousoptimization SOTA Performance achieves state-of-the-art performance byintegrating perception, reasoning,generation Unification Enhances proving unification enhances rather thancompromises specialist strengths From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Fragmented AI Models leads to Inefficiency & Limits. Inefficiency & Limits solves Pelican-Unified 1.0. Pelican-Unified 1.0 uses Unified VLM. Unified VLM enables Chain-of-Thought Reasoning. Chain-of-Thought Reasoning allows Simultaneous Optimization. Unified VLM leads to SOTA Performance. Chain-of-Thought Reasoning leads to SOTA Performance. SOTA Performance shows Unification Enhances solves uses enables allows leads to leads to shows Fragmented AIModels disparatespecialized modelsfor perception,… Inefficiency &Limits leads toinefficiencies andlimits holistic… Pelican-Unified1.0 first unifiedembodied foundationmodel built on… Unified VLM singlevisual-languagemodel maps diverse… Chain-of-ThoughtReasoning autoregressivereasoning generatestask- and… SimultaneousOptimization backpropagation oflosses into sharedrepresentation… SOTA Performance achievesstate-of-the-artperformance by… UnificationEnhances proving unificationenhances ratherthan compromises… From startuphub.ai · The publishers behind this format

The pursuit of truly intelligent embodied agents has long been hampered by the need to train disparate, specialized models for perception, reasoning, and action. This fragmentation leads to inefficiencies and limits the holistic capabilities of AI systems. The introduction of Pelican-Unified 1.0 marks a significant departure, presenting the first embodied foundation model built on the principle of unification.

Unifying Perception, Reasoning, and Imagination

Pelican-Unified 1.0 leverages a single Visual-Language Model (VLM) to serve as a unified understanding and reasoning module. This VLM maps diverse inputs, scenes, instructions, visual contexts, and action histories, into a shared semantic space. Crucially, it also performs autoregressive chain-of-thought reasoning, generating task- and action-oriented sequences in a single pass. This unified approach allows for the backpropagation of language, video, and action losses into the shared representation, enabling simultaneous optimization of understanding, reasoning, imagination, and action, rather than relying on isolated expert systems.

Specialist Strength Without Compromise

Contrary to the intuition that unification might lead to diluted capabilities, Pelican-Unified 1.0 demonstrates that this paradigm can preserve and even enhance specialist performance. A single checkpoint of the model achieved impressive results across multiple domains: 64.7 on eight VLM benchmarks (outperforming comparable-scale models), a first-place ranking of 66.03 on WorldArena, and 93.5 on RoboTwin (second-best among action methods). These findings underscore the efficacy of the unified approach in consolidating complex AI capabilities without sacrificing individual performance.

© 2026 StartupHub.ai. All rights reserved. Do not enter, scrape, copy, reproduce, or republish this article in whole or in part. Use as input to AI training, fine-tuning, retrieval-augmented generation, or any machine-learning system is prohibited without written license. Substantially-similar derivative works will be pursued to the fullest extent of applicable copyright, database, and computer-misuse laws. See our terms.
#AI Research#Embodied AI#Foundation Models#Robotics#Machine Learning