MiniMax review

On pay as you go rates, input costs $0.30 and output costs $1.20. Both sit above the site median for this category, so a chat product with heavy token use should model the bill before settling on a model. The OpenAI-compatible API keeps a trial cheap to run.

On input price from, MiniMax sits 59 of 71 providers, and on output price from 60 of 70. Both places sit well down the field, so the rate itself, not the rank, is the reason to look at a cheaper gateway.

The largest window across the MiniMax models is 1,000,000 tokens, in line with the site median and well above the lower quartile for this category. Long documents and large code bases fit in a single call, and the figure is the maximum across the catalogue rather than a typical size.

The catalogue lists MiniMax M3, MiniMax M2.7 and MiniMax M2.7 highspeed. That covers chat and reasoning text models, speech, image generation and music, so voice features can ship on the same account as chat traffic. One API style covers the whole list, which keeps a mixed product on a single integration.

Yes, the API is OpenAI-compatible, so chat formatting, streaming and tool calls written for OpenAI can be repointed with configuration changes rather than a rewrite. The same endpoint family covers the catalogue, MiniMax M3, MiniMax M2.7 and MiniMax M2.7 highspeed, which mixes text, speech, image and music models.

Support runs on email and community, so there is no live chat in the mix. Urgent problems go by email, and community channels suit background questions on models and billing. Teams that need an instant response should weigh that against the breadth of MiniMax M3, MiniMax M2.7 and MiniMax M2.7 highspeed.

Salad's gateway lists input at $0.06, far below MiniMax's $0.30, which matters for token heavy workloads. MiniMax answers with range instead: MiniMax M3, MiniMax M2.7 and MiniMax M2.7 highspeed covers text, speech, image and music, alongside a 1,000,000 context window.

Teams that need chat text, speech, image and music from one vendor, reached through an OpenAI-compatible API with a 1,000,000 context window. It is a weaker fit for buyers shopping only on entry rate, where Salad's $0.06 input price wins, or for teams after a third party catalogue such as gpt 6 astra, gpt 6 luna and gpt 6 sol.