MiniMax M2.5 API pricing

6 providers serve MiniMax M2.5. MiniMax M2.5 is an open-weight model (Modified MIT) with 229B total parameters, 10B active per token (mixture-of-experts), up to 1000K context. Prices below are per 1M tokens, cheapest input first (self-hosting math further down).

Refreshed 2026-08-23 · list prices from provider APIs

Input floor $0.300/1M at AWS Bedrock, Baseten, MiniMax and 3 more Output floor $1.10/1M at OpenRouter Find your workload winner →

Provider	Input $/1M	Output $/1M	Cache read $/1M	Context
AWS Bedrock	$0.300	$1.20	-	1000K	Use
Baseten	$0.300	$1.20	-	-	Use
MiniMax	$0.300	$1.20	$0.030	1000K	Use
OpenRouter	$0.300	$1.10output floor	$0.150	197K	Use
W&B Inference	$0.300	$1.20	-	197K	Use
Tensormesh	$0.300	$1.20	-	197K

The cheapest way to run MiniMax M2.5

On an illustrative 70M-input / 30M-output monthly workload, today's lowest listed cost is roughly $54.00/month, at OpenRouter. 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 (~$0.137/1M 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: MiniMax M2.5 self-hosting economics

MiniMax M2.5 is open-weight (Modified MIT), so the API price above competes with the GPU-hour market. 229B total parameters (MoE, ~10B active per token) needs roughly 315 GB VRAM at FP8 or 158 GB at INT4, KV-cache headroom included.

No single GPU fits MiniMax M2.5 at FP8 (315 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 158 GB. Tune it in the calculator.

Breakeven estimate: a well-batched vLLM deployment on 2× MI300X ($1.00/hr) at ~4050 aggregate tok/s and 50% utilization works out to roughly $0.137 per 1M output tokens, versus $1.10 via the cheapest output-token API (OpenRouter). 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 M2.1	$0.270	$1.20	6
MiniMax M2	$0.255	$1.02	6
MiniMax M2.7	$0.240	$0.550	4
MiniMax M3	$0.300	$1.20	3
MiniMax M3 (batch)	$0.300	$1.20	1

FAQ

What is the cheapest API for MiniMax M2.5?

As of 2026-08-23, the lowest input price for MiniMax M2.5, $0.300 per 1M tokens, is the same list price at AWS Bedrock, Baseten, MiniMax and 3 more. The lowest output price is OpenRouter at $1.10 per 1M output tokens. The cheapest provider for a real workload depends on its input/output mix.

How much VRAM do you need to self-host MiniMax M2.5?

MiniMax M2.5 has 229B parameters, so plan for roughly 315 GB of VRAM at FP8 or 158 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