Who has the fastest Qwen3.5 397B-A17B inference?

Fireworks AI serves Qwen3.5 397B-A17B at 140 tokens/sec. Serverless endpoint. For latency-sensitive workloads this is usually the right pick even if not the cheapest.

Is the output quality identical across all providers hosting Qwen3.5 397B-A17B?

The weights are identical · Apache 2.0. Differences come from quantization (some providers use int8 or fp8 for speed), context window caps, and provider-added safety filters.

What are cheaper alternative models to Qwen3.5 397B-A17B?

See our substitute finder for models within 10% performance at lower price.

Arbitrage · Qwen3.5 397B-A17B5 providers · $0.280 → $0.600 · 53% spread

Cheapest Provider for Qwen3.5 397B-A17B

Alibaba's flagship MoE model · 397B total / 17B active · multilingual leader.

397B (17B active)Apache 2.0Model detail page →

Cheapest input

Alibaba Cloud

$0.280/M

Direct from Alibaba · 1M context

Fastest

Fireworks AI

140 tok/s

Serverless endpoint

Savings calculator

Save 53%

vs DeepInfra at $0.600/M input. For 100M tokens/mo, that is $32/mo saved by routing to Alibaba Cloud.

Sorted by input price

All 5 providers

Provider	In $/M	Out $/M	Context	Speed	Free	Region
Alibaba CloudWinner	$0.280	$0.850	1.0M	65 t/s		CN/APAC
H Hyperbolic	$0.350	$0.350	128K	80 t/s		US
Together AI	$0.450	$0.450	128K	110 t/s		US
Fireworks AI	$0.500	$1.50	128K	140 t/s		US
D DeepInfra	$0.600	$1.80	33K	60 t/s	—	US

Notes: Alibaba Cloud · Direct from Alibaba · 1M context ; Hyperbolic · Upstart · aggressive pricing ; Together AI · MoE-optimized · flat output ; Fireworks AI · Serverless endpoint ; DeepInfra · Lower context cap

Frequently Asked Questions

Alibaba Cloud at $0.280/M input and $0.850/M output. That is 53% cheaper than DeepInfra. Direct from Alibaba · 1M context.

Cheapest Provider for Qwen3.5 397B-A17B

All 5 providers

Models cheaper than Qwen3.5 397B-A17B

Frequently Asked Questions

This model

Arbitrage

Explore