Qwen: Qwen2.5 7B Instruct
qwen/qwen-2.5-7b-instruct
Cheapest provider
$0.04 / 1M
AtlasCloud
Fastest provider (p95)
123 tok/s
Together
Intelligence (estimated)
60.5
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 |
|---|---|---|---|---|---|---|---|
| AtlasCloud | q8· fp8 | $0.0400 | $0.1000 | — | — | — | 0.0% |
| Phala | undisclosed | $0.0400 | $0.1000 | 522ms / 1145ms | 59 / 88 tok/s | 99.97% | 100.0% |
| Together | q8· fp8 | $0.3000 | $0.3000 | 325ms / 783ms | 75 / 123 tok/s | 99.59% | 100.0% |
“—” 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 sm family base
- ·v2.5 generation +3
Want a curated score? File a benchmark report and we’ll add it to the next sync.
How Atlas routes Qwen: Qwen2.5 7B Instruct
- 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: Qwen2.5 7B Instruct that’s currently AtlasCloud at $0.04/1M.
- Batch tier — async, biggest discount. Roadmapped to use provider batch APIs (OpenAI / Anthropic) where available and queued spot capacity for open-weight workloads.