Qwen 3.5 Omni Plus
Qwen 3.5 Omni Plus is Alibaba Cloud's next-generation native multimodal (Omni) flagship model. Powered by an end-to-end unified architecture, it supports real-time full-duplex interaction across text, image, audio, and video inputs and outputs. It features ultra-low latency streaming inference, expressive speech synthesis, joint vision-audio reasoning, and strong tool-use capability—ideal for real-time voice assistants, AI customer service, video interaction, and digital human agents.
Related Models
Qwen Image 3.0
Qwen Image 3.0 is Alibaba's next-generation open-weights AI image generation and editing model built on a Diffusion Transformer (DiT) architecture. Supporting both text-to-image (T2I) and image-to-image instruction editing (I2I), it excels at rich content rendering, hyper-realistic detail, and deep contextual understanding. It achieves industry-leading text-in-image typography, rendering clear small text down to 10px across 12 native languages and 20+ fonts for complex multi-column layouts, UI designs, and graphic documents. With open weights available for self-hosting, Qwen Image 3.0 provides highly cost-effective, high-precision visual generation for localized advertising, product UI prototyping, e-commerce graphics, and creative publishing.
Qwen Image 3.0 Pro
Qwen Image 3.0 Pro is Alibaba's flagship third-generation AI image generation model built on a Diffusion Transformer architecture. Engineered for high information density and professional production, it accepts ultra-long prompts of up to 4,500 tokens, allowing a single pass to generate complex layouts like newspaper front pages, multi-panel storyboards, academic papers, and detailed UI interfaces. It delivers micro-level detail and industry-leading typography, rendering legible small text down to 10px across 12 native languages and 20+ fonts.
Qwen 3.8 Max
Qwen3.8-Max is Alibaba’s flagship model in the Qwen3.8 series, designed for agentic, text-based workflows. It excels at coding, debugging, office automation, productivity tasks, tool use, and long-horizon autonomous execution. With a 1-million-token context window and support for outputs of up to 64K tokens, it is ideal for processing large documents, repository-scale coding, multi-step planning, structured content generation, and complex workflows requiring sustained reasoning across hundreds or even thousands of steps.
Qwen 3.7 Plus
Qwen3.7-Plus is a cost-effective model in Alibaba’s Qwen3.7 series, supporting text and image inputs with text output. It combines enhanced vision-language capabilities with full-stack, agentic intelligence for coding, tool use, and productivity workflows. Its key strength is multimodal, interactive agent functionality: it can understand real-world scenes, interpret on-screen content, interact with graphical interfaces, generate code from visual references, and autonomously navigate mobile apps from end to end.
Read Me
Qwen 3.5 Omni Plus API
Qwen 3.5 Omni Plus is a omnimodal large language model released by Alibaba on March 30, 2026, as the high-quality variant of the Qwen3.5-Omni series. It supports four input modalities — text, image, audio, and video — with text output and a context window of 262,144 tokens (262K). Alibaba positions it as a comprehensive evolution of Qwen3-Omni, supporting over 10 hours of audio understanding and over 400 seconds of 720p (1 FPS) audio-video understanding and conversation, covering 60+ languages for audio input and 30+ languages for speech output.
On iCreat, you can call qwen3.5-omni-plus through an OpenAI Chat Completions–compatible API. The endpoint supports standard system and user messages, multimodal content arrays, streaming output, and function calling — letting applications already using the OpenAI SDK integrate directly.
Model Positioning
Qwen 3.5 Omni Plus extends the Qwen series from text+vision to full omnimodal understanding, covering long audio understanding, audio-video conversation, multilingual speech interaction, and multimodal content analysis.
For teams needing to process text, image, audio, and video in a single model, it is a practical choice. Qwen 3.5 Omni Plus supports reasoning and function calling, but does not support structured output or context caching; applications requiring these should choose Qwen3.7-Plus or Qwen3.8-Max. Through iCreat's OpenAI-compatible endpoint, the model outputs text only (modalities: ["text"]); native speech output requires Alibaba Cloud Bailian's native API.
Core Capabilities
Omnimodal Understanding
Qwen 3.5 Omni Plus processes text, image, audio, and video inputs simultaneously, correlating cross-modal information within a single request. Supports over 10 hours of continuous audio input and over 400 seconds of 720p (1 FPS) video input, suited for long audio transcription, video content analysis, and multimedia understanding.
Multilingual Speech Interaction
Supports 60+ languages and dialects for audio input recognition, 30+ languages for speech output. Covers a wide range from Mandarin and English to dialects like Minnan, suitable for cross-lingual voice assistants and multilingual content processing.
Reasoning
The Plus variant supports reasoning, performing internal inference before answering — suited for complex multimodal tasks requiring multi-step reasoning. The Flash variant does not support reasoning.
Function Calling
Supports function calling to connect the model with external tools and systems, suited for agent orchestration and tool-use workflows.
Web Search
Supports web search, allowing the model to answer based on real-time retrieved data — suited for queries requiring up-to-date information.
OpenAI Compatibility
Fully compatible with the OpenAI Chat Completions API. Standard parameters such as temperature, max_tokens, and top_p are supported — no additional adaptation needed. Can be accessed directly via OpenAI Python/Node.js SDK through iCreat.
Pricing
| Token Type | Price |
|---|---|
| Input | $7.25 / million tokens |
| Output | $5.50 / million tokens |
This is the unified billing price on the iCreat platform, covering all input modalities (text, image, video, and audio). Alibaba Cloud Bailian's native API uses tiered pricing (audio input and text/image/video input are priced separately); for per-modality billing, refer to Alibaba Cloud Bailian's official pricing.
Use Cases
- Long Audio Understanding & Transcription: 10+ hours of continuous audio input for meeting transcription, podcast analysis, and interview summarization.
- Audio-Video Conversation & Interaction: 400+ seconds of 720p audio-video understanding for video-based Q&A and dialogue.
- Multilingual Voice Assistant: 60+ languages for audio input, 30+ languages for speech output — build cross-lingual voice interaction applications.
- Multimodal Content Analysis: Input video, audio, image, and text simultaneously for cross-modal comprehensive analysis.
- Agent Workflows: Supports function calling and web search for agent orchestration and multi-step task coordination.
- Multimedia Content Creation: Generate text descriptions, summaries, and structured content from image, video, and audio materials.
Model Comparison
Qwen 3.5 Omni Plus vs Qwen 3.5 Omni Flash
| Dimension | Qwen 3.5 Omni Plus | Qwen 3.5 Omni Flash |
|---|---|---|
| Positioning | High-quality omnimodal, with reasoning | Lightweight omnimodal, cost-efficient |
| Context Window | 262,144 tokens | 262,144 tokens |
| Max Output | 65,536 tokens | 65,536 tokens |
| Input Modalities | Text/Image/Video/Audio | Text/Image/Video/Audio |
| Reasoning | Supported | Not supported |
| Function Calling | Supported | Supported |
| Web Search | Supported | Supported |
| Official Input Price | $0.40 / M tokens | $0.10 / M tokens |
| Official Output Price | $4.80 / M tokens | $0.80 / M tokens |
| Release Date | Mar 30, 2026 | Mar 30, 2026 |
| Best For | Reasoning-grade multimodal tasks | High-throughput multimodal workloads |
Qwen 3.5 Omni Plus vs Qwen 3.7 Plus, Qwen 3.8 Max
| Dimension | Qwen 3.5 Omni Plus | Qwen 3.7 Plus | Qwen 3.8 Max |
|---|---|---|---|
| Positioning | Omnimodal (incl. audio/video) | Multimodal agent (text/image/video) | Flagship base (text/image/video) |
| Release Date | Mar 30, 2026 | Jun 1, 2026 | Aug 3, 2026 |
| Context Window | 262,144 tokens | 1,048,576 tokens (1M) | 1,048,576 tokens (1M) |
| Max Output | 65,536 tokens | 65,536 tokens | 131,072 tokens |
| Input Modalities | Text, Image, Video, Audio | Text, Image, Video | Text, Image, Video |
| Audio Input | Supported (10+ hours) | Not supported | Not supported |
| Reasoning | Supported | Supported (deep reasoning) | Supported |
| Function Calling | Supported | Supported | Supported |
| Structured Output | Not supported | Supported | Supported |
| Context Caching | Not supported | Supported | Supported (implicit cache) |
| Official Input Price | $0.40 / M tokens | ~$0.21 / M tokens | $2.00 / M tokens |
| Official Output Price | $4.80 / M tokens | ~$0.63 / M tokens | $6.00 / M tokens |
| Best For | Audio-video understanding & multilingual speech | Multimodal agent & tool use | Coding, office & long-horizon tasks |
Why Choose Qwen 3.5 Omni Plus?
Choose Qwen 3.5 Omni Plus when workloads require audio understanding, video analysis, or multilingual speech interaction — it is the only model in Alibaba's Qwen series covering full omnimodal input (including audio). 10+ hours of audio input, 400+ seconds of video understanding, 60+ language audio recognition, and 30+ language speech output make it suited as the reasoning core for voice assistant, multimedia analysis, and cross-lingual interaction applications.
Compared to Qwen3.7 Plus and Qwen3.8 Max, 3.5 Omni Plus has a smaller context window (262K vs 1M) and lacks structured output, but it is the only model supporting audio input and speech output. If the workload does not involve audio, Qwen3.7 Plus or Qwen3.8 Max are stronger in context length and text capabilities.
Through iCreat, teams can integrate via an OpenAI-compatible endpoint without deploying Alibaba Cloud Bailian's native SDK. Playground is suited for prompt-level testing; for tool integration, state management, and multi-turn conversation, use the API to build production workflows.
Specifications
| Category | Details |
|---|---|
| Model Name | Qwen 3.5 Omni Plus |
| Developer | Alibaba |
| Model ID | qwen3.5-omni-plus |
| Release Date | March 30, 2026 |
| Model Type | Omnimodal Large Language Model |
| Architecture | Thinker-Talker + Hybrid-Attention MoE, ~30B total / ~3B active parameters |
| Context Window | 262,144 tokens (262K) |
| Max Input | 196,608 tokens (196K) |
| Max Output | 65,536 tokens (65K) |
| Input Modalities | Text, Image, Video, Audio |
| Output Modalities | Text, Audio (iCreat endpoint supports text only) |
| Audio Input | 10+ hours continuous audio, 60+ languages and dialects |
| Video Input | 400+ seconds 720p (1 FPS), 113 languages for speech recognition |
| Speech Output | 30+ languages, 55 voice timbres (requires native API) |
| Reasoning | Supported |
| Function Calling | Supported |
| Structured Output | Not supported |
| Web Search | Supported |
| Context Caching | Not supported |
| iCreat API Capabilities | OpenAI Chat Completions compatible, streaming, function calling |
| Primary Tasks | Audio-video understanding, multilingual speech, multimodal analysis, agent workflows |
Architecture
Qwen 3.5 Omni Plus belongs to Alibaba's Qwen3.5-Omni series, featuring a Thinker-Talker architecture with a Hybrid-Attention Mixture-of-Experts (MoE) backbone. The Thinker module ingests text, audio, image, and video inputs and performs reasoning; the Talker module generates text and streaming speech output. The model has approximately 30B total parameters with ~3B active parameters per token, achieving efficient inference through sparse activation.
Unlike Qwen3.7 Plus and Qwen3.8 Max, which use a text+vision multimodal architecture, 3.5 Omni Plus integrates audio encoders and speech synthesis decoders at the architecture level for end-to-end omnimodal processing. This design ensures that text reasoning performance (MMLU-Redux 94.2) remains nearly on par with the non-omnimodal Qwen3.5-Plus (94.3) — adding native audio and video capabilities did not significantly degrade text reasoning.
Notes
When calling through iCreat's OpenAI-compatible endpoint, modalities supports only ["text"], meaning text output only. For speech output, use Alibaba Cloud Bailian's native API with output_mode: text_audio.
The 262,144-token context is a shared request space. Audio and video inputs consume significant tokens (10 hours of audio or 400 seconds of video can fill most of the window) — reserve space for system instructions, conversation state, and the final answer.
Multimodal inputs require accessible URLs; direct local file upload is not supported. The input_audio.format field must match the actual audio format. Common supported formats include wav, mp3, flac, and ogg. Video URLs must point to actual video files (e.g., .mp4), not image URLs.
This model does not support context caching or structured output. For cache-optimized cost or JSON Schema-constrained output, choose Qwen3.7 Plus or Qwen3.8 Max.
Evaluate based on complete task performance. Measure multimodal understanding accuracy, audio recognition WER (word error rate), video understanding accuracy, reasoning quality, latency, and token usage.
FAQ
What's the difference between Qwen 3.5 Omni Plus and Qwen 3.5 Omni Flash?
Plus supports reasoning; Flash does not. Both support full omnimodal input (text/image/video/audio) and function calling. Official pricing: Plus input $0.40/M, output $4.80/M; Flash input $0.10/M, output $0.80/M. Choose Plus for complex tasks requiring reasoning; choose Flash for high-throughput lightweight tasks.
Can I use the OpenAI SDK directly?
Yes. Set base_url to https://api.icreat.ai/llm/openai/v1 and api_key to your iCreat API Key — no additional adaptation needed. Standard parameters like temperature, max_tokens, and top_p are all supported.
How do I enable streaming output?
Add "stream": true to the request body. The response is returned chunk by chunk in SSE (Server-Sent Events) format. Each chunk contains delta.content; the last chunk has finish_reason set to stop.
Can I output speech through iCreat?
No. Through iCreat's OpenAI-compatible endpoint, modalities supports only ["text"], producing text output only. Qwen 3.5 Omni Plus natively supports speech output (30+ languages, 55 voice timbres), but this requires Alibaba Cloud Bailian's native API.
How does it compare to Qwen 3.8 Max?
Qwen 3.8 Max is Alibaba's flagship model released August 3, 2026, with 2.4T parameters, 1M context, and 131K max output — stronger in coding and office capabilities. However, it does not support audio input or speech output. Choose 3.5 Omni Plus for audio-video understanding or multilingual speech; choose 3.8 Max for ultra-long context, coding, or complex text reasoning.
What audio formats are supported?
The input_audio.format field specifies the audio format. Common supported formats include wav, mp3, flac, and ogg. This field must match the actual audio format. Audio data is provided via the input_audio.data field as a URL.
How long can audio and video inputs be?
Supports over 10 hours of continuous audio input and over 400 seconds of 720p (1 FPS) video input. These inputs consume significant tokens — be mindful of the context window budget.