Qwen3 235B A22B API pricing

10 providers serve Qwen3 235B A22B. Qwen3 235B A22B is an open-weight model (Apache-2.0) with 235B total parameters, 22B active per token (mixture-of-experts), up to 262K context. Prices below are per 1M tokens, cheapest input first (self-hosting math further down).

Refreshed 2026-10-08 · list prices from provider APIs

Input floor $0.180/1M at DeepInfra Output floor $0.540/1M at DeepInfra Find your workload winner →

Provider	Input $/1M	Output $/1M	Cache read $/1M	Context
DeepInfra	$0.180	$0.540output floor	-	41K	Use
Nebius	$0.200	$0.600	-	262K	Use
Novita AI	$0.200	$0.800	-	41K	Use
Google Vertex AI	$0.220	$0.880	-	262K	Use
Fireworks AI	$0.220	$0.880	-	131K	Use
Replicate	$0.264	$1.06	-	-	Use
OpenRouter	$0.455	$1.82	-	131K	Use
Scaleway	$0.750	$2.25	-	256K	Use
Hyperbolic	$2.00	$2.00	-	131K	Use
Crusoe	$3.00	$3.00	-	262K	Use

The cheapest way to run Qwen3 235B A22B

On an illustrative 70M-input / 30M-output monthly workload, today's lowest listed cost is roughly $28.80/month, at DeepInfra. Input-heavy, output-heavy and cache-heavy workloads can produce different winners. Use the live mix calculator below rather than combining floors from two different providers. On output-token cost alone, a busy self-hosted deployment can work out cheaper (~$0.302/1M output on 2× MI300X). See the math below. Run your own numbers in the breakeven calculator, or read the breakeven math.

Your actual monthly cost

Uses today's list prices

Workload preset Input tokens / month (millions) Output tokens / month (millions) Input served from cache %

Cache read uses the listed cache rate where available; otherwise it falls back to normal input price. Batch, write-cache, volume and negotiated discounts are excluded.

Or run it yourself: Qwen3 235B A22B self-hosting economics

Qwen3 235B A22B is open-weight (Apache-2.0), so the API price above competes with the GPU-hour market. 235B total parameters (MoE, ~22B active per token) needs roughly 324 GB VRAM at FP8 or 162 GB at INT4, KV-cache headroom included.

No single GPU fits Qwen3 235B A22B at FP8 (324 GB needed), so it needs a multi-GPU node. Cheapest tracked option: 2× MI300X (192 GB each) at roughly $1.00/hr total (RunPod). Rent 2× MI300X at RunPod → INT4 quantization drops the requirement to 162 GB. Tune it in the calculator.

Breakeven estimate: a well-batched vLLM deployment on 2× MI300X ($1.00/hr) at ~1841 aggregate tok/s and 50% utilization works out to roughly $0.302 per 1M output tokens, versus $0.540 via the cheapest output-token API (DeepInfra). Self-hosting wins on cost if you can keep the GPU busy. Planning estimate: throughput varies with hardware, quantization, batch size and context. Tune it in the calculator.

Related models

Model	Cheapest in $/1M	Cheapest out $/1M	Providers
Qwen3 Next 80B A3B Instruct	$0.090	$0.900	10
Qwen3 Next 80B A3B Thinking	$0.140	$0.900	10
Qwen3.8 Max (0902)	$1.65	$4.95	9
Qwen3.6 35B A3B	$0.050	$0.400	9
Qwen3 30B A3B	$0.051	$0.200	8
Qwen3 VL 235B A22B Instruct	$0.200	$0.880	8

FAQ

What is the cheapest API for Qwen3 235B A22B?

As of 2026-10-08, the lowest input price for Qwen3 235B A22B is DeepInfra at $0.180 per 1M input tokens. The lowest output price is DeepInfra at $0.540 per 1M output tokens. The cheapest provider for a real workload depends on its input/output mix. The next-lowest input price is Nebius at $0.200, a 11% difference.

How much VRAM do you need to self-host Qwen3 235B A22B?

Qwen3 235B A22B has 235B parameters, so plan for roughly 324 GB of VRAM at FP8 or 162 GB at INT4/AWQ, KV-cache headroom included. That exceeds a single GPU. A typical node is 2× MI300X.

← All models · GPU rental prices · Breakeven calculator · Get this data as JSON