Models
minimax
minimax logo

MiniMax H3 Max

MiniMax H3 Max is a next-generation general-purpose multimodal video model ranking #1 for overall quality, prompt understanding, and aesthetics in all evaluations, while generating a 5-second video in under 3 seconds.

video generationvisiontextimagevideo

Overview

Type
video
Context
—
Max output
—
Input / output
text, image / video

API

Model identifier

minimax/minimax-h3-max

Providers and live performance

Rolling Gateway metrics. Latency is time to first token; throughput is output tokens per second.

P50 / P95 over the last hour
ProviderStatusLatencyThroughputUptime 15m / 1h / 1dInputOutputCache read / writeDiscountRegionsZDRNo training
minimaxOperational——100.00% / 100.00% / 100.00%$0.00/M$0.00/M— / ————No

Pricing

video duration pricing
{ "resolution": "480p", "cost_per_second": "0.05" }, { "resolution": "768p", "cost_per_second": "0.08" }, { "cost_per_second": "0.08" }

Capabilities

video generationvisiontextimagevideo

Supported parameters

No details published for this model.

Reasoning options

options
{ "type": "budget_tokens" }

Video capabilities

supported operations
text-to-video, image-to-video, first-last-frame
supported resolutions
480p, 768p
supported aspect ratios
21:9, 16:9, 4:3, 1:1, 3:4, 9:16
supported durations seconds
5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
supported fps
24
max sample count
1
input limits
{ "image": { "max_count": 2, "max_file_size_mb": 30, "supported_formats": [ "jpg", "jpeg", "png", "webp", "heic", "heif" ], "supported_sources": [ "url" ], "min_dimension_pixels": 256, "max_dimension_pixels": 5760, "min_aspect_ratio": "2:5", "max_aspect_ratio": "5:2" } }

Metadata

type
video
context window
—
max output
—
released
Aug 27, 2026
zero data retention
none
no training
none
api specifications
v2, v3, v4
Yast logo

Self-improving agents that just work.

Describe what you need. Yast builds the agent, connects your tools, runs it on autopilot, and it gets smarter every time.

Get Started