Qwen 3 235B-A22B
Qwen 3 flagship MoE. 235B total / 22B active per token, with built-in 'thinking' and 'non-thinking' modes that trade speed for reasoning depth at inference time. Best open-weight reasoning model for many tasks.
Our verdict
Positioning
Qwen 3 235B-A22B is the early-2025 frontier MoE that made the "frontier model on consumer-tier hardware" pitch credible. 235B total parameters with 22B active per token. The large open-weight Qwen 3.5 model is Qwen3.5-397B-A17B, a different size, so it is not a same-hardware swap.
Strengths
- Apache 2.0 licensed open weights — same permissive license as Qwen 3.5. Commercial use allowed, no MAU clauses, no Llama-style restrictions. Critical for many team deployments.
- Strong reasoning + coding combo at the frontier tier. Qwen's release post reports competitive benchmark results in coding, math and general capabilities against DeepSeek-R1, o1, o3-mini, Grok-3 and Gemini-2.5-Pro.
- Excellent multilingual support — 100+ languages and dialects per the model card.
- MoE architecture has 22B activated per token out of 235B total.
Limitations
- Qwen has a newer checkpoint. Qwen3-235B-A22B-Instruct-2507 is the updated non-thinking version of this model.
- Memory still substantial. ~142 GB at Q4. No 24-GB-card path. A Mac Studio M3 Ultra with 256 GB handles Q4 + 32K context. The 128-GB tier cannot hold the Q4 weights.
- Training data cutoff is older. Q3 was trained earlier — some 2025 events / APIs / library versions aren't reflected.
Should you run this locally?
Yes, if you already have Q3 235B-A22B running in production and the quality is acceptable for your workload.
Probably not, for anyone running a single consumer GPU — the Q4 weights alone are about 142 GB, so it needs more than 128 GB of unified memory or workstation-tier hardware.
Probably not, for anyone whose primary workload is coding (Qwen 2.5 Coder 32B is a coding-focused model that fits 24 GB).
How it compares
- vs DeepSeek V4 Pro (1.6T MoE) → V4 Pro has higher quality ceiling, but at 1.6T parameters its Q4 weights alone would be about 970 GB. Qwen 3 235B-A22B Q4 is about 142 GB. Both have permissive licenses (V4 Pro is MIT, Qwen 3 is Apache 2.0). For accessibility, Qwen 3 wins.
- vs Llama 4 Scout (Meta MoE) → similar quality tier. Meta states up to 10M tokens of context for Llama 4 Scout; Qwen 3 235B-A22B has 32,768 natively and 131,072 with YaRN. Llama license has 700M MAU clause; Qwen 3 has Apache 2.0. Pick on context-length needs + license preference.
- vs DeepSeek R1 (671B reasoning specialist) → R1 specializes in reasoning, Q3 is generalist. For reasoning-only workloads R1 wins; for mixed daily-driver tasks Q3 235B-A22B is more useful.
- vs Qwen 3 30B-A3B (smaller MoE sibling) → 30B-A3B fits 24 GB consumer card — dramatically more accessible. Quality is meaningfully lower. Pick the smaller MoE for "I want Qwen on a 4090"; pick 235B-A22B for "I have workstation hardware and want frontier."
Run this yourself
# Mac Studio M3 Ultra 256GB: Q4 fits
ollama pull qwen3:235b-a22b-q4_K_M
ollama run qwen3:235b-a22b-q4_K_M
# Or via llama.cpp directly:
llama-server -m qwen3-235b-a22b-Q4_K_M.gguf \
--ctx-size 32768 -ngl 999 --temp 0.7
Quant: Q4_K_M GGUF
Context: 32768 (KV cache f16, ~6.3 GB additional)
Backend: llama.cpp Metal via Ollama
Hardware: Mac Studio M3 Ultra 256 GB unified memory
Overview
Qwen 3 flagship MoE. 235B total / 22B active per token, with built-in 'thinking' and 'non-thinking' modes that trade speed for reasoning depth at inference time. Best open-weight reasoning model for many tasks.
Family & lineage
How this model relates to others in its lineage. Family members share architecture and training-data roots; parent / children edges record direct distillation or fine-tune relationships.
Strengths
- Switchable thinking mode
- Apache 2.0
- Top-tier reasoning
Weaknesses
- Needs 160GB+ VRAM at Q4
- Multi-GPU only on consumer rigs
Quantization variants
Each quantization trades model quality for file size and VRAM. Q4_K_M is the most popular starting point.
| Quantization | File size | VRAM required |
|---|---|---|
| Q4_K_M | 142.0 GB | 160 GB |
| Q5_K_M | 167.0 GB | 190 GB |
At Q4 Qwen 3 235B-A22B needs about 160 GB of memory by our estimate, more than a 32 GB graphics card holds. Renting a GPU with enough memory (or several GPUs) by the hour is the practical way to try it before you plan hardware for it.
Referral links: RunLocalAI may earn a commission or account credits if you sign up and rent, at no extra cost to you. Hourly rates and which GPUs are listed change all the time, so check the provider's page before you start. How we make money.
Get the model
Ollama
One-line install
ollama run qwen3:235bRead our Ollama review →HuggingFace
Original weights
Source repository — direct quantization required.
Hardware that runs this
Cards with enough VRAM for at least one quantization of Qwen 3 235B-A22B.
Models worth comparing
Same parameter band, plus what's one tier above and below — so you can decide what actually fits your hardware.
Frequently asked
What's the minimum VRAM to run Qwen 3 235B-A22B?
Can I use Qwen 3 235B-A22B commercially?
What's the context length of Qwen 3 235B-A22B?
How do I install Qwen 3 235B-A22B with Ollama?
Compare against other models
Curated head-to-head decisions where Qwen 3 235B-A22B is one of the contenders. For arbitrary pairings use /model-battle.
Source: huggingface.co/Qwen/Qwen3-235B-A22B
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify model claims.
Related — keep moving
Verify Qwen 3 235B-A22B runs on your specific hardware before committing money.