Qwen3 Coder 30B A3B qwen3-coder-30b-a3b
qwen · qwen coding efficient
#79 overall #14 agentic
- params
- 30.5B (A3.3B)
- arch
- moe
- context
- 256k
- license
- apache-2.0 open weights
- released
- Jul 2025
- downloads/30d
- 745.0k
# architecture
- layers
- 48
- d_model
- 2048
- heads
- 32q / 4kv
- head dim
- 128
- experts
- top-8 of 128
- vocab
- 151936
- family
- qwen3_moe
signal path of one layer, generated from the registry's structured fields — dimension lines quote the model's real numbers; an MoE trace forks at the router, a dense trace runs straight through. Geometry fields come from the repo's config.json.
# why ranked
Overall: #79 — score 28.6 ● high — all signals present
| signal | weight | input (0–100) |
|---|---|---|
| aa_intelligence | 0.70 | 30.1 |
| bench_composite | 0.30 | 25.1 |
benchmark panel evidence:
| complete panel | score (0–100) |
|---|---|
| aa-math-v1 | 5.6 |
| swe-bench-verified-v1 | 44.6 |
Coding: provisional — score 42.1 ● low — a single signal; treat with caution
provisional: too few independent signals for a numbered position — the score is shown, but this model sorts after every ranked model.
| signal | weight | input (0–100) |
|---|---|---|
| aa_coding | 0.30 | — |
| aider_polyglot | 0.30 | — |
| swe_bench_verified | 0.40 | 42.1 |
missing signals are dropped and the remaining weights renormalized — never imputed.
Agentic: #14 — score 42.1 ● low — a single signal; treat with caution
| signal | weight | input (0–100) |
|---|---|---|
| swe_bench_verified | 0.60 | 42.1 |
| swe_rebench | 0.40 | — |
missing signals are dropped and the remaining weights renormalized — never imputed.
# trend
methodology changed during this history window; score movement across that boundary is not model movement. See methodology v6.
# benchmarks
| benchmark | score | source | date |
|---|---|---|---|
| AA Intelligence Index | 9.6 | Artificial Analysis | — |
| AA Math Index | 29.0 | Artificial Analysis | — |
| SWE-bench Verified | 60.4 | SWE-bench | Sep 2025 |
# where to run
| provider | quant | ctx | $/M in | $/M out | $/M cache | price src | tps | uptime | ✓ |
|---|---|---|---|---|---|---|---|---|---|
| Novita | 160k | $0.07 | $0.27 | — | via litellm | — | — | ||
| SiliconFlow | fp8 | 256k | $0.07 | $0.28 | — | openrouter | — | 99.2% | ✓ |
| Amazon Bedrock | unknown | 256k | $0.07 | $0.30 | — | bedrock | — | 100.0% | ✓ |
| Scaleway | 125k | $0.20 | $0.80 | — | via litellm | — | — | ||
| Alibaba | 256k | $0.45 | $2.25 | — | via modelsdev | — | — |
sorted by blended price ((3·input + output) / 4 per 1M) · ✓ = the provider's own catalog confirms the offer · "via …" prices are what the aggregator routing the offer charges, not the provider's own list price
source aliases
- aa
qwen3-coder-30b-a3b-instruct- hf
unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF- openrouter
qwen/qwen3-coder-30b-a3b-instruct- swebench
Qwen3-Coder-30B-A3B-Instruct