Qwen3 Coder 30B A3B Instruct
Qwen · released 2025-07-31 · apache-2.0 license
Qwen3-Coder-30B-A3B-Instruct is a 30.5B (3.3B active) open-weight MoE coding model (July 2025, Apache-2.0) focused on agentic coding with 262K context. Artificial Analysis scores it 20 on its Intelligence Index, above average for its size.
Key specs
| Type | Local open-weight |
|---|---|
| Parameters | 30.53B total · MoE, — active |
| Architecture | qwen3_moe |
| Context window | 262K tokens |
| Knowledge cutoff | 2025-06-30 |
| Modalities | text |
| Recommended backends | — |
| Minimum viable rig | — |
Benchmark scores
| GPQA Diamond | — |
|---|---|
| SWE-bench Verified | — |
| AIME | — |
| MMLU-Pro | — |
| BFCL v3 (tool use) | — |
| Composite score | — |
| Community rating | No reviews yet |
VRAM & disk per quantization
| Quant | VRAM | Disk | RAM | Context |
|---|---|---|---|---|
| Q4_K_M | 19.2 GB | 17.7 GB | — | 262K |
Strengths & weaknesses
Strengths: Agentic coding and tool use for a small MoE (3.3B active); Long context (262K native, up to 1M with YaRN); Above-average intelligence vs similar-size open models
Weaknesses: No official per-eval benchmark table for this size; Non-thinking only