Qwen/Qwen3.5-35B-A3B-FP8
Parameters36B (3B active)
Context256K
Input → outputimage + text → text
Licenseapache-2.0
PrecisionF8_E4M3 · fp8
ArchitectureQwen3_5MoeForConditionalGeneration
ReleasedFeb 25, 2026
Downloads1.3M
Based on Qwen/Qwen3.5-35B-A3B
Other sizes and formats in the family
| Model | Params | Precision | Downloads |
|---|---|---|---|
| Qwen/Qwen3.5-397B-A17B-GPTQ-Int4gptq | 397B | – | 3K |
| Qwen/Qwen3.5-122B-A10B-FP8fp8 | 125B | F8_E4M3 | 507.5K |
| Qwen/Qwen3.5-122B-A10B | 125B | BF16 | 594.2K |
| Qwen/Qwen3.5-122B-A10B-GPTQ-Int4gptq | 125B | I32 | 336.1K |
| Qwen/Qwen3.5-35B-A3B-GPTQ-Int4gptq | 36B | I32 | 203.5K |
| Qwen/Qwen3.5-27B-FP8fp8 | 27.8B | F8_E4M3 | 168.8K |
| Qwen/Qwen3.5-27B-GPTQ-Int4gptq | 27.8B | I32 | 17.8K |
| Qwen/Qwen3.5-9B | 9.7B | BF16 | 8.3M |
| Qwen/Qwen3.5-9B-Base | 9.7B | BF16 | 304.1K |
| Qwen/Qwen3.5-4B | 4.7B | BF16 | 7.9M |
| Qwen/Qwen3.5-4B-Base | 4.7B | BF16 | 653.9K |
| Qwen/Qwen3.5-2B | 2.3B | BF16 | 3.2M |
| Qwen/Qwen3.5-2B-Base | 2.3B | BF16 | 305.8K |
| Qwen/Qwen3.5-0.8B | 873M | BF16 | 2.6M |
| Qwen/Qwen3.5-0.8B-Base | 873M | BF16 | 479.7K |
This release in the news
No coverage of this exact release yet.
Recent news about Qwen
- Oct 11, 2026 · Opinion / analysisEvery Model That Can Be Run On 10-16GB VRAM Ranked
- Oct 11, 2026 · Opinion / analysisReverse Engineering w/ Local?
- Oct 11, 2026 · Benchmark resultAgents That Own Their Inference — Du'an Lightfoot & Khaja Omer, Akamai Technologies
- Oct 11, 2026 · Opinion / analysisUPDATE: Qwen 3.8 27B 140 tok/s on single RTX 3090 Megakernel: KL divergence 0.0009 vs llama.cpp
- Oct 11, 2026 · Model releaseQwen/Qwen-Image-2.1-Turbo
- Oct 11, 2026 · Opinion / analysisBuilding prompts with LLMs
- Oct 11, 2026 · Opinion / analysis[P] Pecision models that score every allowed label from the logits: Jebadiah v2.1 (27B, 9B), open weights and self-run benchmark results [P]
- Oct 11, 2026 · Opinion / analysisPSA: DeepSeek V4.1 Flash habitually exfiltrates API keys. It is dangerously misaligned and may be hazardous to use