MiniMax Open Platform Review: Pricing & Comparison
MiniMax's official open platform — text, image, video, speech and music all behind one API
Last verified: 2026-08-06 · Visit official site →
One account, five modalities
Most AI vendors specialize in either text chat or one generation modality (image/video/speech). MiniMax has taken a harder road: it trains its own models across text, image, video, speech and music, and bundles all five into a single open platform (platform.minimaxi.com) with one unified account and billing system.
That’s an outlier among the 10 new vendors added in this round — most of the others specialize in one direction (image or video). MiniMax looks more like an “all-modality bundle,” similar in breadth to aggregators like 302.AI or Shenma Relay, except MiniMax is the original model creator, not a relay forwarding third-party models.
Model lineup
- Text: MiniMax-M3, positioned as a “native multimodal, 1M-context Frontier Coding model,” and also one of the more popular domestic open-source text models
- Video: Hailuo H3 (海螺视频), supporting text-to-video, image-to-video, first/last-frame control and multimodal reference, at 768P or 2K resolution with 4-15 second clips
- Image: the image-01 / image-01-live series, supporting text-to-image and image-to-image
- Speech: Speech-2.8-HD / Speech-2.8-Turbo, emphasizing natural sound quality and generation speed
- Music: music-3.0 / music-2.6 / music-cover, supporting composition and vocal-cover generation
The video capability (Hailuo H3) already has a large real user base through the consumer product “Hailuo AI” (hailuoai.com), so it’s been validated in the market, not just showcased in developer docs.
The June 2026 pricing controversy: worth knowing
To be candid, MiniMax went through a notable pricing controversy in mid-2026: the cheapest Token Plan tier jumped from ¥29 to ¥49/month, a more than 65% increase, and users discovered that domestic subscription pricing ran roughly 28% higher than the overseas plan at the time (adjusted for exchange rate). After the backlash, parent company Xiao-i Technology publicly apologized, acknowledging it “failed to communicate the adjustment sufficiently in advance,” and preserved existing users’ prior benefits while issuing some free credit to new users.
This episode doesn’t necessarily reflect a technical or product problem, but pricing stability is a reasonable factor to weigh when evaluating whether to depend on a vendor long-term — check current official pricing before integrating, and build some budget headroom for potential price changes.
Who it’s for
If your product needs both text chat and generation across image/video/speech, and you’d rather not integrate a separate vendor per modality, MiniMax’s “all-modality” positioning can save real integration work. If you only need one modality done at the highest possible quality (say, only the best video generation), a vertical specialist like Kling or Vidu may perform better.
Information verified 2026-08-06. Model lists and pricing reflect the current MiniMax Open Platform website; check the site for any further changes following the June 2026 pricing adjustment.
Related reviews
- Kling AI: Kuaishou’s official Kling open platform, an industry-benchmark video-generation model
- Vidu: Shengshu Technology’s Vidu video-generation model, one of the first domestic Sora-class video models with an open API
- PixVerse: AIsphere’s short-video generation model, a unicorn valued at over $2B
- Zhipu AI: Zhipu’s official GLM-5 flagship, multimodal AI with top-tier Chinese-language capability
Quick facts
| Pricing model | Token Plan monthly subscription from ¥49 up to the ¥119 Max tier covering all modalities; pay-as-you-go text pricing roughly ¥1/M tokens input, ¥8/M tokens output; speech/video also available as lower-priced prepaid resource packs |
|---|---|
| Model coverage | MiniMax-M3 text model (a native multimodal, 1M-context Frontier Coding model), Hailuo H3 video generation (text-to-video/image-to-video/first-last-frame/multimodal reference), the image-01 series for image generation, Speech-2.8 for speech synthesis, and music-3.0 for music generation |
| Latency / SLA | Direct mainland China access via in-house inference clusters; latency varies by modality/model, with no unified published SLA |
| Mainland direct connect | Official domestic China product, no proxy needed |
| Best for | Developers / Enterprise |
| Referral program | No public affiliate program found so far. |
Pros
- A genuine 'one API for every modality' offering: text (MiniMax-M3), video (Hailuo H3), image (image-01), speech (Speech-2.8) and music (music-3.0) all live under the same open-platform account, so you don't need a separate vendor per modality
- MiniMax-M3 is itself one of the more popular domestic open-source text models — if your product needs both text chat and multimodal generation, this can save you a separate LLM-vendor integration
- Video generation (Hailuo) has accumulated a large real-user base through the consumer app 'Hailuo AI,' so it's been validated on the consumer side, not just a developer-facing experimental capability
Cons
- A June 2026 Token Plan price increase drew user backlash — the cheapest subscription tier rose from ¥29/month to ¥49/month, and users found domestic pricing running roughly 28% higher than the overseas plan; the company publicly apologized and offered compensation afterward, so it's worth watching pricing stability going forward
- The domestic platform (platform.minimaxi.com) and the international platform (platform.minimax.io) are two separate account systems with non-interchangeable billing currencies and plans, so cross-region teams need to register and manage both separately
- As an 'all-modality' platform, the depth of any single modality (say, pure image generation) may not match vertical specialists like Fal AI or Kling
Compare more AI API relays
See the full comparison board — filter by price tier, model coverage, and mainland direct-connect status.
Back to the comparison board →