midjourney · Rate Limits

Midjourney Rate Limits

Midjourney does not currently expose a public, metered HTTP API. Generations are scoped to Discord and the web app and are governed primarily by per-plan Fast GPU minute budgets and concurrent-job caps rather than per-second rate limits. Concurrency limits and Relax-mode queue behavior vary by tier.

Midjourney Rate Limits is the machine-readable rate-limit profile for midjourney on the APIs.io network, conforming to the API Commons Rate Limits specification.

It captures 3 rate-limit definitions, measuring gpu-minute, concurrent_requests, and varies.

The profile also includes 3 backoff/retry policies defined.

Tagged areas include Rate Limiting, AI, and Image Generation.

3 Limits
Rate LimitingAIImage Generation

Limits

Fast GPU minutes (monthly) account
gpu-minute · month
Per-plan; see plan page
Concurrent jobs (Fast) account
concurrent_requests
Per-plan; typically 3 (Basic) up to 12+ (Mega) — verify on live docs
Relax mode queue account
varies
Standard / Pro / Mega only; queued behind Fast jobs, no per-minute SLA

Policies

Plan-bound concurrency
Each tier has a fixed maximum of concurrent Fast jobs; additional generations queue until a slot frees.
Relax mode degradation
Once Fast minutes are exhausted, supported tiers fall back to Relax mode (queued) until the next billing cycle or Fast top-up.
Stealth gating
Stealth Mode is a Pro / Mega entitlement, not a rate limit, but affects which generations are publicly visible.

Sources