MiniMax M3 API pricing

8 providers serve MiniMax M3. MiniMax M3 is an open-weight model (MiniMax Community) with 428B total parameters, 23B active per token (mixture-of-experts), up to 1049K context. Prices below are per 1M tokens, cheapest input first (self-hosting math further down).

Refreshed 2026-08-28 · list prices from provider APIs

Input floor $0.230/1M at W&B Inference Output floor $0.960/1M at W&B Inference Find your workload winner →

Provider	Input $/1M	Output $/1M	Cache read $/1M	Context
W&B Inference	$0.230	$0.960output floor	$0.050	262K	Use
DeepInfra	$0.280	$1.10	$0.056	524K	Use
Fireworks AI	$0.300	$1.20	$0.060	512K	Use
MiniMax	$0.300	$1.20	$0.060	1000K	Use
Together AI	$0.300	$1.20	$0.060	524K	Use
Tencent	$0.300	$1.20	$0.060	1000K
Novita AI	$0.300	$1.20	$0.060	1000K	Use
OpenRouter marketplace quote	$0.300	$1.20	$0.060	1049K	Use

The cheapest way to run MiniMax M3

On an illustrative 70M-input / 30M-output monthly workload, today's lowest listed cost is roughly $44.90/month, at W&B Inference. 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 (~$0.631/1M output on 4× 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: MiniMax M3 self-hosting economics

MiniMax M3 is open-weight (MiniMax Community), so the API price above competes with the GPU-hour market. 428B total parameters (MoE, ~23B active per token) needs roughly 589 GB VRAM at FP8 or 295 GB at INT4, KV-cache headroom included.

No single GPU fits MiniMax M3 at FP8 (589 GB needed), so it needs a multi-GPU node. Cheapest tracked option: 4× MI300X (192 GB each) at roughly $2.00/hr total (RunPod). Rent 4× MI300X at RunPod → INT4 quantization drops the requirement to 295 GB. Tune it in the calculator.

Breakeven estimate: a well-batched vLLM deployment on 4× MI300X ($2.00/hr) at ~1761 aggregate tok/s and 50% utilization works out to roughly $0.631 per 1M output tokens, versus $0.960 via the cheapest output-token API (W&B Inference). 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
MiniMax M2.5	$0.300	$1.10	7
MiniMax M2.1	$0.270	$1.20	6
MiniMax M2.7	$0.250	$0.550	6
MiniMax M2	$0.255	$1.02	6
MiniMax M3 (batch)	$0.300	$1.20	1

FAQ

What is the cheapest API for MiniMax M3?

As of 2026-08-28, the lowest input price for MiniMax M3 is W&B Inference at $0.230 per 1M input tokens. The lowest output price is W&B Inference at $0.960 per 1M output tokens. The cheapest provider for a real workload depends on its input/output mix. The next-lowest input price is DeepInfra at $0.280, a 22% difference.

How much VRAM do you need to self-host MiniMax M3?

MiniMax M3 has 428B parameters, so plan for roughly 589 GB of VRAM at FP8 or 295 GB at INT4/AWQ, KV-cache headroom included. That exceeds a single GPU. A typical node is 4× MI300X.

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