Product guide

Find the right AI for you.

Start from a task,
explore tools, models and integration.

Multimodal Interaction Model Platform

Whatever you want to do, find it here.

Six categories, 84 features. Find tools by task, then view quality-processing results in the original media.

Browse the feature catalog

Berserk

Understand the problem. Generate the content.

Glacier AI's in-house large model family. From chat to image and video generation, explore model capabilities by task.

Explore in-house models

AI Aggregator

Many models. One integration.

Starting from your business systems, confirm model scope, protocols, keys and project permissions for a clearer integration path.

Learn about model integration

From needs to capabilities.

Expand a category to see full feature descriptions.

VideoGeneration, super-resolution & editing15 features

Text-to-video and image-to-video

Generate video clips from text prompts or reference images for creative content, ad assets and previsualization.

Video super-resolution and frame interpolation

Increase video resolution and frame rate for smoother, sharper footage; suited to restoring old footage and slow motion.

Video highlight detection and auto-editing

Automatically identify highlights and edit them into short clips for sports, livestreams and short-video production.

Video subtitle removal

Detect and remove burned-in subtitles for re-dubbing and multilingual distribution.

Video watermark removal

Locate and remove video watermarks/logos for footage cleanup and compliance.

Video stabilization

Remove handheld shake for steady footage and a better viewing experience.

Video matting and background replacement

Separate foreground people from the background, supporting virtual backgrounds as a green-screen alternative.

Action and event recognition

Recognize human actions and behavior events in video for security, sports and industrial monitoring.

Activity, fall and fatigue detection

Detect falls, fatigue and abnormal activity with alerts for elder care, construction sites and driving.

Visual anomaly detection

Identify deviations from normal patterns in video for production lines, inspections and security.

Multi-object tracking

Continuously track multiple targets in video and output trajectory IDs for pedestrian and traffic flow analysis.

Object detection and multi-object tracking

Detects objects in video and images and tracks them continuously, supporting counting, speed measurement and behavior analysis.

Trajectory prediction and motion planning

Predicts the future trajectories of pedestrians, vehicles and robots for autonomous driving and robot obstacle avoidance.

Lip sync and digital human presenting

Aligns a digital human's lip movements with speech for virtual anchors, broadcasts and customer service.

Vision-language-action and grasping

Combines vision and language instructions to drive robotic grasping for embodied AI.

Browse video capabilities
TextWriting, translation & extraction16 features

General writing and chat

Text generation, rewriting, Q&A and multi-turn chat for general assistant scenarios.

Summarization and rewriting

Condenses long text, rewrites style and extracts key points for reports, news and meeting minutes.

Machine translation and localization

Translation between multiple languages with localization polish, for documents, interfaces and subtitles.

Text moderation

Identifies violating, violent or terrorist, sexual and advertising text for communities and content platforms.

Text embeddings and semantic similarity

Encodes text as vectors for semantic search, deduplication, clustering and similarity matching.

Entity and relation extraction

Extracts people, organizations, events and their relations from text to build knowledge graphs.

Sentiment and opinion analysis

Determines text sentiment polarity and opinion for public opinion monitoring, reviews and customer service QA.

Intent recognition and slot extraction

Understands user intent and extracts key slots for dialogue systems and ticket routing.

Table classification and risk scoring

Classifies tables and forms and outputs risk scores for risk control, compliance and review.

Field and form extraction

Extracts structured fields from contracts, receipts and forms to automate data entry.

Text language identification

Automatically detects the language of text for routing, pre-translation and content distribution.

Language identification

Identifies the language of speech or text for multilingual customer service and meeting systems.

PII/PHI detection and redaction

Detects and redacts personally identifiable and health information for privacy compliance.

AI safety detection and response

Identifies risks such as prompt injection, jailbreaks and harmful content, and triggers response policies.

Causal inference and uplift modeling

Estimates the causal effects and incremental gains of interventions for marketing, operations and pricing.

Graph, anti-fraud and relationship risk

Mines graphs for fraud rings and linked risks for anti-fraud and risk control.

Browse text capabilities
ImageGeneration, restoration & recognition16 features

Text-to-image and reference-image generation

Generate images from text or reference images for creative work, design and marketing assets.

Inpainting, outpainting and controlled editing

Local redrawing, outpainting and controlled editing for photo retouching, e-commerce and design.

Image classification and fine-grained recognition

Identifies image categories and fine-grained subcategories for quality inspection, archiving and recognition.

OCR text recognition

Recognize text in images, including receipts, IDs, documents and scene text.

Layout, table and formula parsing

Parses document layout, table structure and math formulas for document digitization.

Face detection, recognition and liveness

Face detection, matching and liveness checks for access control, identity verification and security.

Pose, gesture and keypoints

Recognizes body poses, hand gestures and keypoints for interaction, fitness and animation.

Medical image detection & segmentation

Detect and segment lesions and organs to support diagnosis and research.

Supervised defect detection

Detect surface and internal product defects from labeled samples for industrial quality inspection.

Visual measurement & alignment

Measure dimensions, position and alignment for precision manufacturing and assembly.

Open-vocabulary detection & segmentation

Detect and segment targets described in natural language, with no fixed categories.

Image & video segmentation

Semantic, instance and panoptic segmentation for cutouts, labeling and analysis.

3D reconstruction & digital twins

Reconstruct 3D scenes from images or point clouds for digital twins, mapping and simulation.

Depth estimation

Estimate monocular or stereo depth for autonomous driving, AR and robotics.

Super-resolution, denoising and deblurring

Image enhancement and restoration for quality improvement and old photo restoration.

Text/image to 3D assets

Generate 3D models and assets from text or images for games, e-commerce and the metaverse.

Browse image capabilities
SpeechTranscription, synthesis & translation13 features

Speech-to-text ASR

Transcribe speech to text for meetings, customer service and subtitles.

Text-to-speech (TTS)

Synthesize text into natural speech for announcements, audio content and accessibility.

Real-time voice support & business agent

Real-time voice interaction that performs business actions, for call centers and voice assistants.

Speech-to-speech translation

End-to-end speech translation that preserves timbre and prosody for cross-language communication.

Voice cloning & conversion

Clone voices or convert speaking styles for dubbing and personalized TTS.

Speaker diarization

Distinguish between speakers for meeting minutes and multi-speaker transcription.

Voiceprint recognition & verification

Confirm identity by voiceprint for identity verification and anti-fraud.

Wake word detection

Detect a specified wake word for low-power voice assistant wake-up.

Voice activity detection (VAD)

Detect speech and non-speech segments to cut costs, split audio and support real-time interaction.

Denoising & speech enhancement

Suppress noise and enhance voices for calls, recordings and meetings.

Echo cancellation & dereverberation

Remove echo and reverberation for hands-free calls and conference room audio.

Environmental sound recognition

Identify sound events in the environment for security, industry and urban sensing.

Music & sound effect generation

Generate music and sound effects for content creation, games and short videos.

Browse speech capabilities
Cross-modalImage-text understanding & knowledge Q&A14 features

RAG knowledge Q&A

Retrieval-augmented generation that answers from a company knowledge base and reduces hallucinations.

Document Q&A & comparison

Q&A, comparison and summarization across documents such as contracts, reports and papers.

Enterprise knowledge assistant

An all-in-one assistant for internal knowledge search, Q&A and writing.

Multimodal retrieval

Cross-modal retrieval such as text-to-image/video and image-to-text search.

Visual Q&A & scene understanding

Answer questions about images for accessibility, education and inspection.

Real-time multimodal assistant

An interactive assistant that processes speech, images and text in real time.

Tool calling & workflow agent

Call external tools and APIs to orchestrate workflows and complete complex tasks.

GUI & Browser Agents

Operate graphical interfaces and browsers to complete tasks, for automation and RPA.

Coding & DevOps Agents

Code generation, comprehension, review, and ops automation.

Code Generation & Comprehension

Code completion, generation, explanation, refactoring, and unit test generation.

Multi-Agent Orchestration for Research & Ops

Multi-agent collaboration for research, operations, and analysis tasks.

Personalized Ranking

Personalized ranking based on user behavior, for recommendations, search, and ads.

Candidate Retrieval & Similar Recommendations

Retrieve candidates and recommend similar items, for e-commerce, content, and social.

Semantic Search & Rerank

Semantic retrieval + reranking for more relevant search.

Browse cross-modal capabilities
Time series & sensingTime series forecasting & scene sensing10 features

Predictive Maintenance & Remaining Useful Life

Predict equipment failures and remaining useful life from multi-source data.

Streaming Anomaly & Change Detection

Real-time detection of anomalies and sudden changes in streaming data.

Time Series Forecasting

Trend, seasonality, and anomaly forecasting for time series data.

Scheduling, Routing & Resource Optimization

Solve scheduling, routing, and resource allocation optimization problems.

AMR Navigation & Dispatch

Mobile robot navigation, obstacle avoidance, and multi-robot dispatch.

Drone & Autonomous Site Inspection

Autonomous drone inspection, recognition, and alerts.

Multi-Sensor Perception & Fusion

Fused perception across cameras, radar, LiDAR, and other sensors.

IMU/Radar/LiDAR Fusion

Fuse inertial, radar, and LiDAR data for localization and perception.

Visual-Inertial SLAM

Visual + inertial odometry and mapping, for robotics and AR.

Load & Environmental Forecasting

Forecast power load and environmental metrics, for energy and city management.

Browse time series & sensing capabilities

This is a capability catalog. Specific available models, usage and integration settings are confirmed based on your task.

More resources, whenever you need them.

View text and structured materials, or open the original media to see processing results.

Berserk models and pricing

The chat models listed in current product materials include Berserk-V4flash, Berserk-5.2 and Berserk-H3, referencing DeepSeek V4 flash, GLM5.2 and MiniMax H3 respectively. A reference direction does not mean a performance comparison has been completed; actual performance and available versions are confirmed per project.

Under current product materials, pricing is 50% of the lowest market price that day for the corresponding reference model. This describes a product rule, not a real-time market quote; reference prices may fluctuate, and actual billing and unit prices are subject to the project quotation.

AI Aggregator model coverage

The model catalog covers OpenAI, Anthropic, Google, DeepSeek, Qwen, Kimi, Zhipu, MiniMax, xAI, Kuaishou and more.View all models and prices to see how each model is billed: by token, per call or by duration.

Bring your ideas,
and take the next step.

Tell us about your task, existing systems and goals, and we'll identify the right capabilities together.

GLACIER AI

Start with your needs.

Tell us which products interest you, your existing systems and your goals.

shenzhongchang@glacier.fit

Product scope, configuration and activation are subject to our discussion.