Qwen3 30B A3B Instruct 2507 qwen3-30b-a3b-instruct-2507
qwen · qwen
unranked — no quality signals yet
- params
- 30.5B
- arch
- moe
- context
- 256k
- license
- apache-2.0 open weights
- released
- Jul 2025
- downloads/30d
- 1.1M
# 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.
# trend
collecting history — trends appear after a few daily runs.
# where to run
| provider | quant | ctx | $/M in | $/M out | $/M cache | price src | tps | uptime | ✓ |
|---|---|---|---|---|---|---|---|---|---|
| DekaLLM | unknown | 256k | $0.09 | $0.30 | — | via openrouter | — | 100.0% | |
| SiliconFlow | fp8 | 256k | $0.09 | $0.30 | — | openrouter | — | 97.0% | ✓ |
| Wandb | 262k | $0.10 | $0.30 | — | via litellm | — | — | ||
| Nebius | fp8 | 256k | $0.10 | $0.30 | — | openrouter | — | 100.0% | ✓ |
| Alibaba | 128k | $0.20 | $0.80 | $0.20 | via requesty | — | — |
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
- openrouter
qwen/qwen3-30b-a3b-instruct-2507