Qwen: Qwen3.6 27B
qwen/qwen3.6-27b
Cheapest provider
$0.29 / 1M
Io Net
Fastest provider (p95)
—
No throughput data yet — populated as traffic accumulates
Intelligence (estimated)
71.0
Family + generation + popularity + price tier
Per-provider performance
Latency / throughput / uptime measured across providers over the last 30 minutes of live traffic. Atlas’s router weighs these per call (with the eval-gate signal) when picking a variant for Standard and Batch tiers.
| Provider | Quant | Input $/1M | Output $/1M | Latency p50 / p95 | Throughput p50 / p95 | Uptime 30m | Success |
|---|---|---|---|---|---|---|---|
| Io Net | q8· fp8 | $0.2900 | $3.2000 | — | — | 98.88% | — |
| Chutes | q8· fp8 | $0.3000 | $2.0000 | — | — | 98.66% | — |
| SiliconFlow | q8· fp8 | $0.3000 | $3.2000 | — | — | 97.01% | — |
| Ambient | undisclosed | $0.3200 | $3.2000 | — | — | 99.00% | — |
| DeepInfra | q8· fp8 | $0.3200 | $3.2000 | — | — | 100.00% | — |
| Venice | q8· fp8 | $0.3250 | $3.2500 | — | — | 100.00% | — |
| Alibaba | undisclosed | $0.4500 | $2.7000 | — | — | 100.00% | — |
| Morph | undisclosed | $0.4980 | $2.4000 | — | — | 98.18% | — |
| WandB | q8· fp8 | $0.6000 | $3.6000 | — | — | 100.00% | — |
“—” means live telemetry hasn’t accumulated enough recent traffic for that endpoint. “undisclosed” means the provider serves the model but doesn’t expose the quantization label (typically running fp8 / int8 internally).
Intelligence estimate
No public benchmark numbers indexed for this model yet, so the leaderboard score is derived from the catalogue data we sync: model family standing, generation, and live usage rank — applied identically to open- and closed-weight models. The heuristic ceiling sits below confirmed-benchmark frontier models so curated rankings stay clearly on top.
- ·qwen lg family base
Want a curated score? File a benchmark report and we’ll add it to the next sync.
How Atlas routes Qwen: Qwen3.6 27B
- Realtime tier — direct passthrough at the upstream’s native precision. Best for hard latency / quality guarantees.
- Standard tier — Atlas picks the cheapest provider variant whose quantization has stayed green on your operation’s eval gates. For Qwen: Qwen3.6 27B that’s currently Io Net at $0.29/1M.
- Batch tier — async, biggest discount. Roadmapped to use provider batch APIs (OpenAI / Anthropic) where available and queued spot capacity for open-weight workloads.