Qwen3 32B qwen3-32b
qwen · qwen hybrid-reasoning
unranked — no quality signals yet
- params
- 32.8B
- arch
- dense
- context
- 128k
- license
- apache-2.0 open weights
- released
- Apr 2025
- reasoning
- yes
# architecture
- layers
- 64
- d_model
- 5120
- heads
- 64q / 8kv
- head dim
- 128
- vocab
- 151936
- family
- qwen3
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: provisional — score 28.3 ● 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_intelligence | 0.70 | — |
| bench_composite | 0.30 | 28.3 |
benchmark panel evidence:
| complete panel | score (0–100) |
|---|---|
| aider-polyglot-v1 | 30.1 |
| aime-2025-v1 | 26.5 |
missing signals are dropped and the remaining weights renormalized — never imputed.
Coding: provisional — score 36.4 ● 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 | 36.4 |
| swe_bench_verified | 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
# where to run
| provider | quant | ctx | $/M in | $/M out | $/M cache | price src | tps | uptime | ✓ |
|---|---|---|---|---|---|---|---|---|---|
| Featherless | — | — | — | — | via hfrouter | — | — | ||
| OVHcloud | 32k | $0.08 | $0.23 | — | via litellm | — | — | ||
| Nscale | 40k | $0.08 | $0.25 | — | via hfrouter | 46 | — | ||
| DeepInfra | fp8 | 40k | $0.08 | $0.28 | — | deepinfra | — | 98.4% | ✓ |
| Amazon Bedrock | — | $0.07 | $0.30 | — | bedrock | — | — | ✓ | |
| Nebius | 125k | $0.10 | $0.30 | $0.10 | via requesty | — | — | ||
| Chutes | fp8 | 40k | $0.10 | $0.42 | — | chutes | — | — | ✓ |
| Tenstorrent | fp8 | 20k | $0.14 | $0.40 | $0.09 | via openrouter | — | 99.9% | |
| Phala | 40k | $0.12 | $0.50 | — | phala | — | — | ✓ | |
| SiliconFlow | fp8 | 128k | $0.14 | $0.57 | — | openrouter | — | 96.5% | ✓ |
| Bedrock Mantle | 32k | $0.15 | $0.60 | — | via litellm | — | — | ||
| Alibaba | 128k | $0.70 | $2.80 | — | 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
qwen3-32b- aider
Qwen3 32B- arena
Qwen3-32B- openrouter
qwen/qwen3-32b