- Context
- 131Ktokens
- Released
- 1y ago
- Size
- 235Bparameters
Qwen3-235B-A22B is the flagship of the Qwen3 series, the latest addition to Alibaba's Qwen family of large language models, released with open weights. Its launch announcement, titled 'Think Deeper, Act Faster,' highlights a hybrid design in which the model can either reason step by step on complex problems or answer directly for speed, with the reasoning effort set per task; Alibaba says it achieves competitive results against top-tier models on coding, math and general capabilities. It is built for coding and agentic work, with strengthened MCP support and coverage of 119 languages.
- Language model
- Open weights · 235B
- Reasoning
- Tools
How it compares
Size class
Large235BAn open-weight model of 100 billion parameters or more, which needs a multi-GPU server.
Open models of a similar size from other labs
Vs. your stack
Add the models you use to see how this one compares.
Sign inAdditional source evidence
- Source model
- alibaba/qwen3-235b-a22b
- Source family
- qwen
- Last updated
- 2025-04-28
- Knowledge cutoff
- 2025-04
- Model context
- 131K
- Model output limit
- 16K
- Open weights
- Yes
- File attachments
- No
- Reasoning
- Yes
- Tool calling
- Yes
- Temperature control
- Yes
Downloadable weights
Published benchmarks
Complete source record (JSON)
{
"id": "alibaba/qwen3-235b-a22b",
"name": "Qwen3 235B-A22B",
"description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use",
"family": "qwen",
"last_updated": "2025-04-28",
"knowledge": "2025-04",
"attachment": false,
"reasoning": true,
"tool_call": true,
"temperature": true,
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"limit": {
"context": 131072,
"output": 16384
},
"benchmarks": [
{
"name": "Aider Polyglot",
"score": 59.6,
"source": "https://aider.chat/docs/leaderboards/",
"metric": "percent correct",
"date": "2025-05-09"
},
{
"name": "SWE-Bench Pro",
"score": 21.41,
"source": "https://labs.scale.com/leaderboard/swe_bench_pro_public",
"metric": "resolve rate",
"dataset": "public"
}
],
"release_date": "2025-04-28",
"open_weights": true,
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/Qwen/Qwen3-235B-A22B"
}
]
}Frequently asked questions
Running locally
Estimates are 164.3 GB at Q4, 287.6 GB at Q8 and 541 GB at FP16 for the 235B parameter model. All VRAM figures are estimates: weights plus about 15% and 0.5 GB for a short context.
No. Qwen3 235B-A22B does not fit on a 24 GB, 16 GB, 12 GB or 8 GB GPU at Q4, Q8 or FP16 (estimate).
Yes, at Q4 (estimate). Estimated VRAM at Q4 is 164.3 GB, and assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits is 256 GB.
Privacy and data
Yes. Qwen3 235B-A22B has open weights, so you can self-host it for full control of your data.
Capabilities and specs
The context window is 131K tokens (131,072) and the maximum output is 16K tokens (16,384).
No. The input modalities of Qwen3 235B-A22B are text only.
Yes, tool calling (function calling) is supported.
Yes, a reasoning mode is supported.
The knowledge cutoff of Qwen3 235B-A22B is 2025-04.
Quality and benchmarks
Qwen3 235B-A22B ranks 164 on Arena Text (Arena's own rank) with a score of 1402.
The closest alternatives, matched on model kind and release date (no comparable price), are FastVLM 0.5B by Apple, FastVLM 1.5B by Apple and Llama 3.1 Nemotron 70B Instruct by NVIDIA.
FastVLM 0.5BQwen3 235B-A22B vs FastVLM 0.5BFastVLM 1.5BQwen3 235B-A22B vs FastVLM 1.5BLlama 3.1 Nemotron 70B InstructQwen3 235B-A22B vs Llama 3.1 Nemotron 70B Instruct
Versions and lineage
Qwen3 235B-A22B has a context of 131K tokens (131,072) versus 33K tokens (32,768) for Qwen3 8B. Qwen3 235B-A22B was released April 28, 2025, and Qwen3 8B on April 29, 2025.
Licensing and access
Qwen3 235B-A22B is open weights.
Basics
Qwen released Qwen3 235B-A22B on April 28, 2025.
Yes, Qwen3 235B-A22B has open weights.
Generated from Vertical Tabs data, updated November 11, 2026.
Discussion
No comments yet
Be the first to share your thoughts.