Qwen3-VL-235B-A22B
Frontier-level multimodal intelligence for complex visual reasoning, GUI automation, and sophisticated coding.
Total Parameters
235B
22B Activated
Context Window
262K
Tokens
Modalities
Text, Image
& Video
Max Output
16,384
Tokens
Frontier-Level Vision & Language Unification
Qwen3-VL-235B-A22B Instruct is a massive open-weight multimodal model that unifies strong text generation with deep visual understanding. Delivering performance comparable to GPT-5 Chat and Claude 4 Opus Thinking, it excels at advanced reasoning, complex code generation, and 2D/3D spatial grounding. Whether you need multilingual document parsing, complex multi-image dialogue, or visual coding workflows, Qwen3-VL provides maximum intelligence for your most demanding applications where quality is paramount.
Vision-Language MoE — 22B active / 235B total
Built for visual intelligence
Advanced Visual Perception
Excels at general vision-language tasks including VQA, chart/table extraction, document parsing, and robust recognition of real-world and synthetic categories.
Agentic GUI Automation
Operates GUI elements for automation tasks, aligns text to video timelines for precise temporal queries, and follows complex instructions over multi-turn dialogues.
Visual Coding Workflows
Transforms UI mockups and whiteboard sketches directly into functional code and actively assists with UI debugging and spatial reasoning.
Frontier-Level Reasoning
Delivers text-only performance comparable to flagship proprietary models, making it ideal for sophisticated mathematical analysis and complex project execution.
Multimodal Intelligence
Proven performance across reasoning, coding, and agentic workflows.
Overall Intelligence
Better than 48% of models
Coding Capability
Better than 50% of models
Agentic Capability
Better than 49% of models
| Category | Benchmark | Score |
|---|---|---|
| Reasoning | GPQA Diamond | 71.2% |
| Reasoning | τ²-Bench Telecom | 35.1% |
| Reasoning | IFBench | 42.7% |
| Reasoning | AA-LCR | 31.7% |
| Reasoning | GDPval-AA | 8.1% |
| Reasoning | HLE | 6.3% |
| Reasoning | CritPt | 0.0% |
| Coding | SciCode | 35.9% |
| Coding | Terminal-Bench Hard | 6.8% |
| Knowledge | AA-Omniscience | 20.2% |
Metrics sourced from Artificial Analysis.
Flexible Pricing Tiers
Choose the optimal balance of speed and cost for your workflow. Prices are per 1M tokens.
| Tier | Input / 1M tokens | Output / 1M tokens |
|---|---|---|
| Standard | $0.10 | $0.40 |
| Async | $0.15 | $0.55 |
| Realtime | $0.60 | $1.20 |
Context window natively supported up to 262k tokens.
Start Building in Minutes
Qwen3-VL-235B-A22B is accessible via OpenAI-compatible endpoints. Here is how to integrate it using the standard Python SDK via Doubleword.ai.
Developer Tip: Recommended Sampling Parameters
For optimal performance and to reduce endless repetitions, the Qwen team recommends: Temperature=0.7, TopP=0.8, TopK=20, MinP=0, and a presence_penalty=1.5 (adjust up to 2.0 if repetition persists).
from openai import OpenAI
client = OpenAI(
api_key="your-api-key-here",
base_url="https://api.doubleword.ai/v1"
)
# Step 1: Upload a batch input file
with open("batch_requests.jsonl", "rb") as file:
batch_file = client.files.create(
file=file,
purpose="batch"
)
print(f"File ID: {batch_file.id}")
# Step 2: Create a batch job
batch = client.batches.create(
input_file_id=batch_file.id,
endpoint="/v1/chat/completions",
completion_window="24h"
)
print(f"Batch ID: {batch.id}")
# Step 3: Check batch status
batch_status = client.batches.retrieve(batch.id)
print(f"Status: {batch_status.status}")