MiniMax pricing covers three separate products under one brand, which is why it confuses people. The Hailuo AI video generator runs from a free tier up to $199.99/month. The MiniMax API bills its flagship MiniMax M3 at $0.30 input and $1.20 output per million tokens. The developer subscription, now called the M Plan after it replaced the old Token Plan, starts at $22/month. Every figure below was re-checked against MiniMax’s own pages on October 17, 2026.
The naming is the first hurdle. Hailuo AI is MiniMax’s consumer video app, MiniMax is the company and its API brand, and searches like “hailuo minimax pricing” mix the two together. This guide separates all three so you can find your exact plan, see what a single video or a million tokens actually costs, and decide which tier is worth paying for. Prices here move often, so treat every figure as a starting point and confirm at the source before you buy.
The Key Takeaways
- Hailuo AI video plans run from a free tier to $199.99/month (Max), with paid plans listing at $14.99 to $199.99 and a limited-time promo cutting the entry Standard tier to $7.99/month.
- Paid plans include 1,000 to 20,000 monthly credits, enough for roughly 40 to 900 videos at economical settings; credits reset monthly and do not roll over.
- The MiniMax API bills MiniMax M3 and M2.7 at $0.30/$1.20 per million tokens; M3 prompts above 512K tokens cost $0.60/$2.40. OpenRouter’s discount has shrunk to a few cents, with M2.5 at $0.27/$0.95.
- The developer M Plan, which replaced the Token Plan, costs $22/month (Go), $55/month (Explore) or $132/month (Build). Annual billing gives 2 months free, and only Explore and Build include the H3 video model.
- All paid Hailuo plans remove the watermark, add commercial use rights, and bundle MiniMax’s own H3 video model with other labs’ models like Veo 3.1 and Nano Banana Pro. The plan page still lists Sora 2, but OpenAI has shut Sora down.
How Much Does MiniMax Cost? The Short Answer
MiniMax sells three things, and each is priced on its own logic. The Hailuo AI video generator uses monthly credit subscriptions aimed at creators, from free up to $199.99/month. The MiniMax API is pay-as-you-go per token for developers, with MiniMax M3 at $0.30/$1.20 per million tokens on the official sheet. The M Plan, which replaced the Token Plan (also sold as the Coding Plan), is a flat developer subscription from $22/month that swaps metered billing for shared usage windows.
If you just want to make AI videos, you want Hailuo. If you are wiring MiniMax models into your own software, you want the API or the M Plan. The sections below cover all three, with the consumer video plans first since that is what most people searching for “hailuo minimax pricing” actually need. Hailuo also ranks among the best free AI video generators, so you can start without paying anything.
Hailuo AI Pricing: Video Plans and Credits
Hailuo AI runs on credits. Every subscription hands you a monthly credit pool, and each video spends credits based on its resolution, length, and model. A short 768p clip is the cheapest output, while higher resolutions and longer durations burn through credits faster, so the video counts below assume the most economical settings. Credits reset each month and unused ones expire, which matters a lot when you pick a tier.
Here are the current Hailuo plans, re-checked on October 17, 2026 on the official Hailuo subscription page. Prices and credit counts have not changed since June. The list prices below reflect the standard monthly rate; a running limited-time promotion still discounts the three lower tiers, and the page shows the promo price you actually pay at checkout.
| Plan | List / month | Current promo | Monthly credits | Videos/month (approx) |
|---|---|---|---|---|
| Free | $0 | N/A | Trial credits | Trial only |
| Standard | $14.99 | $7.99 | 1,000 | ~40 |
| Pro | $54.99 | $24.99 | 4,500 | ~180 |
| Master | $94.99 | $63.99 | 10,000 | ~400 |
| Ultra | $124.99 | No promo | 12,000 | ~480 |
| Max | $199.99 | No promo | 20,000 | ~900 |
Hailuo Free and Standard ($7.99 to $14.99)
The Free plan is fine for trying things out, but it gives you limited trial credits, keeps the watermark, restricts you to Hailuo’s own models, and sits at the lowest queue priority. The Standard plan lists at $14.99/month and is currently discounted to $7.99 on a limited-time promo that the company says reverts to $14.99 once it ends. It includes 1,000 credits, around 40 six-second 768p videos or about a dozen 1080p clips, removes the watermark, and adds commercial rights, though it caps video length at six seconds and runs tasks one at a time.
Hailuo Pro and Master ($24.99 to $94.99)
The Pro plan lists at $54.99/month, currently $24.99 on promo, and bumps you to 4,500 credits, roughly 180 economical videos, plus ten-second 1080p output and two clips rendering in parallel. The Master plan lists at $94.99/month, currently $63.99, with 10,000 credits and around 400 videos. This is the band most regular creators land on, since 4,500 to 10,000 credits is enough to publish a steady stream of clips without rationing every generation.
Hailuo Ultra and Max ($124.99 to $199.99)
The Ultra plan costs $124.99/month for 12,000 credits, about 480 videos, and adds unlimited use of the older Hailuo 01 model at no credit cost. The Max plan, at $199.99/month for 20,000 credits and roughly 900 videos, is the most comprehensive tier, with unlimited access to both the Hailuo 01 and 02 models plus the agent. Both top tiers keep video output at up to ten-second 1080p; the 2K and 4K “unlimited” perks Hailuo advertises apply to the bundled image models, not the video clips themselves.
How Many Videos Do Hailuo Credits Buy?
The credit-to-video ratio is the number that actually decides value. At the cheapest settings, a six-second 768p clip costs roughly 25 credits, so 1,000 credits yields about 40 videos. Step up to ten-second 768p and you halve that count; jump to six-second 1080p and a single Standard plan stretches to only about 12 clips. That is why the “videos per month” figures swing so widely, and why creators who work mostly in 1080p should size up at least one tier beyond what the headline credit number suggests.
The bigger story is what each paid plan adds beyond MiniMax’s own video engine, which is now H3, alongside the older Hailuo 2.3. Paid tiers bundle access to Veo 3.1 and Seedance 2.0, and from Pro upward you also get image models like Nano Banana Pro, Nano Banana 2, Seedream, and GPT Image. The plan page still lists Sora 2 as well, but OpenAI has shut Sora down, so do not pick a plan for it. That turns Hailuo from a single video tool into a multi-model creative suite. For a broader look at how it stacks up against rivals, see our roundup of the best AI video generators.
MiniMax API Pricing: Per-Token Costs by Model
The MiniMax API is pay-as-you-go, billed per million tokens with no base subscription required. MiniMax’s official sheet charges a flat $0.30/$1.20 for M3, M2.7, and the legacy M2.5, M2.1 and M2. High-speed M2.x variants and M3 prompts above 512K tokens bill at $0.60/$2.40, and M3’s optional Priority tier costs 1.5 times the standard rate in exchange for faster, more reliable responses. The table shows the official rate next to OpenRouter’s, read on October 17, 2026.
| Model | Official /1M (in/out) | OpenRouter /1M (in/out) | Context | Best for |
|---|---|---|---|---|
| MiniMax M3 | $0.30 / $1.20 | $0.30 / $1.20 | 1M | Flagship, long context |
| MiniMax M2.7 | $0.30 / $1.20 | $0.30 / $1.20 | 205K | Current mid-tier |
| MiniMax M2.5 | $0.30 / $1.20 (legacy) | $0.27 / $0.95 | 205K | Cheapest output via OpenRouter |
| MiniMax M2 | $0.30 / $1.20 (legacy) | $0.255 / $1.02 | 205K | Cheapest input via OpenRouter |
| MiniMax M2.1 | $0.30 / $1.20 (legacy) | $0.30 / $1.20 | 205K | Legacy |
| MiniMax-01 | Retired | $0.20 / $1.10 | 4M | Legacy long-context |
| MiniMax M1 | Retired | $0.55 / $2.20 | 1M | Legacy reasoning |
MiniMax M2.5 and M2: The Cheapest Entry on OpenRouter
MiniMax M2.5 is still a strong value pick for coding workloads; we covered it in detail in our MiniMax M2.5 review. MiniMax now files it under legacy models at the flat $0.30 input and $1.20 output per million tokens. Resellers still undercut that, though by far less than before. OpenRouter lists M2.5 at $0.27/$0.95 and the older M2 at $0.255/$1.02, both on a 205K-token context window. That is a saving of roughly 10% to 20%, against the $0.12/$0.48 OpenRouter charged for M2.5 earlier in 2026.
MiniMax M3: The 1M-Context Flagship ($0.30/$1.20)
The flagship MiniMax M3 carries a 1M-token context window and sits at $0.30/$1.20 thanks to a permanent 50% discount off its $0.60/$2.40 list rate; prompts above 512K tokens bill at that higher $0.60/$2.40 tier. The optional Priority tier costs $0.45/$1.80. That makes M3 one of the cheapest million-token-context models you can call, and a natural pick for agents that need to hold large codebases or long documents in a single prompt. If you run agentic coding workloads, another open-weight option worth pricing out is Nex-N2-Pro, a 397B MoE model built for long-horizon coding.
Buying MiniMax Through OpenRouter
You can also buy these models through resellers. OpenRouter MiniMax pricing matches the official rate on M3 and M2.7 at $0.30/$1.20 and shaves a little off the legacy models, listing M2.5 at $0.27/$0.95. It also still carries retired models like MiniMax-01 and M1 that MiniMax has dropped from its own pricing page. Pricing this low keeps MiniMax in the same budget bracket as DeepSeek, far below the flagship models from OpenAI or Anthropic.
MiniMax M Plan Pricing (Formerly the Token Plan)
For developers who would rather pay a flat fee than meter every call, MiniMax sells the M Plan, which has replaced the Token Plan (also marketed as the Coding Plan). One subscription key covers MiniMax’s text, image and audio models, plus video on the upper tiers. It works in MiniMax Code and in any tool that speaks the OpenAI or Anthropic API format, such as Claude Code, Codex and Cursor. Every tier runs the new M3.1 Flash Preview text model.
| Tier | Monthly | Annual | Usage vs Go | Video model |
|---|---|---|---|---|
| Go | $22 | $220 | 1x | Not included |
| Explore | $55 | $550 | 3x | H3 |
| Build | $132 | $1,320 | 7.5x | H3 |
Go, Explore, and Build
Go at $22/month suits light research, writing and small code changes, but it has no video model. Explore at $55/month gives three times Go’s usage and adds the H3 video model; MiniMax pitches it for daily coding and writing. Build at $132/month gives 7.5 times Go’s usage for long-context work, complex coding and heavy multimodal creation. Annual billing gives you 2 months free, so Go works out to roughly $18.33/month on the $220 yearly plan.
How M Plan Usage Limits Work
M Plan usage is one shared pool, not a quota per model. Text, image and audio requests count against a 5-hour window and a weekly window. Video counts only against the weekly window, and one video uses far more than a chat message. Unused usage does not roll over. When you hit a limit, MiniMax draws on any credit packs you bought (1,000 credits cost $1 at pay-as-you-go rates), or you can upgrade or wait for the reset. MiniMax recommends pay-as-you-go, not M Plan, for production traffic. If you are weighing this against rival developer subscriptions, our Grok pricing and Mistral AI pricing breakdowns are useful comparison points.
Is MiniMax Worth It? Value vs DeepSeek, GPT-6, and Veo
On raw cost, MiniMax is still one of the cheapest capable model families you can buy, for video and for the API. It is no longer the cheapest, though. The table below puts its token rates next to the budget and frontier options developers actually compare it against.
| Model | Input /1M | Output /1M | Context | Tier |
|---|---|---|---|---|
| GPT-6 Luna | $0.10 | $0.50 | 1M | Budget |
| MiniMax M2.5 | $0.27 | $0.95 | 205K | Budget (OpenRouter) |
| MiniMax M3 | $0.30 | $1.20 | 1M | Budget flagship |
| DeepSeek V4-Pro | $0.66 / $1.32 | $1.98 / $3.96 | 1M | Budget |
| GPT-6 Sol | $2.00 | $10.00 | 1M | Mid-tier |
| Claude Opus 5.5 | $4.00 | $20.00 | 1M | Frontier |
MiniMax M3 inputs cost $0.30 per million. That is less than half of DeepSeek V4-Pro off-peak and less than a quarter at peak, under a sixth of GPT-6 Sol, and about 13 times less than Claude Opus 5.5. The exception is OpenAI’s own budget model. GPT-6 Luna at $0.10/$0.50 undercuts M3 by two thirds on input with the same 1M-token window, so price alone no longer settles the choice. For the full landscape of how these models rank on quality and price, see our guide to the best AI models. DeepSeek V4-Pro is shown as off-peak / peak: peak runs 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday to Friday, and has cost double the off-peak rate since 16 August 2026.
On video, the real value is the bundling. A single Hailuo plan hands you Veo 3.1 and Seedance 2.0 next to MiniMax’s own H3 and a stack of image models, instead of a separate bill for each vendor. The trade-off is the credit system, which forces you to ration generations and watch your monthly balance, and pricing that has crept upward as those frontier models were added. If you mainly want polished short clips and like having several video engines in one place, it is hard to beat on price.
Skip the Credit Math With One Multi-Model Subscription
Juggling Hailuo credits, an API key, and an M Plan with two usage windows adds up fast, both in money and in mental overhead. If what you actually want is answers from strong models, not a metering dashboard, there is a simpler route. Fello AI puts MiniMax M3 next to Claude, ChatGPT, Gemini, Grok, and DeepSeek in one Mac and iPhone app for one flat subscription. You can try MiniMax on a real task and check its answer against another model, including the budget Chinese models MiniMax competes with. There is no second app to open and no credit pack to top up. For video generation itself you still need Hailuo.
For creators, that removes the constant “do I have enough credits” worry. For everyday users, it means you can switch from a coding question to a research task to a writing job without changing apps or paying three separate bills. You can try it from the App Store and see whether one subscription beats stacking several.
Conclusion
MiniMax pricing is cheap by design, whether you are generating video or calling the API. For video, the Hailuo Standard plan, currently $7.99/month on promo, is the sensible starting point, with Pro the upgrade most regular creators land on. For developers, MiniMax M3 at $0.30/$1.20 per million tokens is the value pick for long-context work, though OpenAI’s GPT-6 Luna now costs less. The $22 M Plan Go tier suits anyone who wants a flat monthly bill and does not need video. Because these prices change often, and the lower Hailuo tiers are running a temporary promo, confirm the current numbers on the official MiniMax pricing page before you subscribe.
FAQ
How much does MiniMax cost?
It depends on the product. Hailuo AI video plans run from free to $199.99/month, the API charges $0.30 per million input tokens and $1.20 per million output tokens for MiniMax M3, and the developer M Plan starts at $22/month.
Is Hailuo AI free?
Yes, there is a free plan with limited trial credits, but it restricts you to Hailuo’s own models, keeps a watermark, and gives you the lowest queue priority. Paid plans list from $14.99/month, with a running promo currently dropping the Standard tier to $7.99.
How many videos can I make on the Hailuo Standard plan?
The Standard plan includes 1,000 credits, good for about 40 six-second 768p videos or roughly a dozen 1080p clips, since higher resolution burns more credits per video.
Do Hailuo credits roll over?
No. Subscription credits are valid for one month and expire at the end of your billing cycle, so unused credits do not carry into the next month.
What is MiniMax M3 pricing?
MiniMax M3 costs $0.30 per million input tokens and $1.20 output under a permanent 50% discount off its $0.60/$2.40 list rate, with prompts above 512K tokens billing at that higher tier and a 1M-token context window.