MiniMax M3.1: Flash Preview Access, API Status, and Price
M3.1-Flash-Preview is available in MiniMax Code for everyday development. Compare its access and API status with the unreleased full M3.1 and today's MiniMax M3 price.
Published 2026-09-27
MiniMax M3.1-Flash-Preview launched in MiniMax Code on September 27, 2026. MiniMax positions it for fast, dependable everyday development, from fixing a bug to building a complete feature. This is access inside MiniMax Code, not an announcement of a generally available, separately priced M3.1 API. The full M3.1 tier and M3 Pro have not been released through a public API. The UI's M3.1-Flash-Preview label is not a verified API request ID.
For an API you can call now, use minimax-m3. On September 27, APIMaster listed 3 active routes, from $0.30 input / $1.20 output per 1M tokens against the marketplace's original $0.60 / $2.40 (about 50% off). View the live M3 discount; availability and the card's current rates take precedence over this snapshot.
| Model ID | Original price (1M input / output) | APIMaster current price | Active routes | Discount | Model card |
|---|---|---|---|---|---|
| M3.1-Flash-Preview (MiniMax Code label; API ID unannounced) | Not published | No public route | 0 | — | Not listed |
minimax-m3 |
$0.60 / $2.40 | From $0.30 / $1.20 | 3 | ~50% off | View discount |
GPT Trial(GPT-6 Astra, GPT-6 Sol, GPT-5.6, and GPT-5.5 available)
Sign up and claim a $20 GPT trial
What actually launched on September 27?
The new selection in MiniMax Code is called M3.1-Flash-Preview. MiniMax's announcement describes a fast, stable model for day-to-day software development: bug fixes, implementation and end-to-end feature work. Its preview suffix matters. It identifies an early, lightweight development option rather than a definitive specification for the full M3.1 family.
Developers with MiniMax Code access can select it in the product and run their own repository tasks. A screenshot of the selector also shows an effort control, but a UI control alone does not establish public API parameters, tier availability, token prices or performance guarantees. Test the actual Code workflow before assuming it matches another provider's thinking modes.
MiniMax also announced a Token Plan quota reset and a September 28–October 7 double daily check-in free quota for M3.1-Flash-Preview, H3 and H3 Max. These are MiniMax Code or plan benefits; they are not an APIMaster discount or a public pay-as-you-go API price. Check MiniMax's current campaign terms for eligibility and expiry.
Is the full MiniMax M3.1 API available?
Not as a documented public release at this check. MiniMax has not provided a standalone stable M3.1 API identifier, per-million-token rate, cache price, context window, maximum output, service limits or published benchmark sheet for the full model. Do not send the display label M3.1-Flash-Preview as an assumed API model value. The same distinction applies to any marketplace listing: a search result is not proof of a working route.
| Product | Where to try it | Status | What remains unknown |
|---|---|---|---|
| M3.1-Flash-Preview | MiniMax Code | Preview available | Public API ID, token price, limits and stable API date |
| Full M3.1 | Not published as a distinct API | Pending | Final model card, capabilities, context and commercial terms |
| M3 Pro | Not released | Roadmap / reported specs | Parameter count, architecture, launch date and price |
MiniMax M3 (minimax-m3) |
APIMaster public marketplace | 3 active routes | Confirm live routing and price before use |
M3.1 Flash Preview vs M3 Pro
They are not interchangeable names. The first is a released preview in MiniMax Code optimized for day-to-day coding. M3 Pro is a future, larger flagship candidate. Community reports have suggested roughly 3 trillion parameters and MSA 2.0 for M3 Pro, but neither belongs in an M3.1 Flash specification. The earlier earnings-call roadmap mentioned M3.1, M3 Pro and H3.1 as separate names; it did not turn a later preview into the final versions of all three.
For a useful evaluation, put the Flash Preview through a real bug fix, a multi-file feature and a failing test suite. Track time to a correct patch, unnecessary edits, tool-call reliability and regressions. Do not use a rumor-driven parameter comparison as a proxy for coding quality.
MiniMax M3 API pricing and a working integration today
The available M3 marketplace card is the practical baseline while M3.1 stays in preview. The lowest active minimax-m3 route currently lists $0.30 input / $1.20 output per 1M tokens; the original price displayed by the marketplace is $0.60 / $2.40. There are 3 active routes as of September 27, 2026. This 50% difference is relative to the card's original rate, not an announced price for M3.1. A route can change or disappear; check its live card before relying on it.
Create an APIMaster account and API key, then send a small test request to the listed M3 model:
curl "https://apimaster.ai/v1/chat/completions" \
-H "Authorization: Bearer $APIMASTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"minimax-m3","messages":[{"role":"user","content":"Explain why this unit test fails."}]}'
This example is for M3, not M3.1 Flash Preview. Confirm model compatibility, response shape and current card terms on your own route before scaling.
What to watch before switching to M3.1
- An official MiniMax model card naming the final release stage and exact API request ID.
- Input, output and cache billing, including plan-specific limits and whether Flash Preview carries over.
- Context window, maximum output, tool use and multimodal input documentation for that exact model.
- A callable route with sufficient uptime for your own repository and agent tasks.
Keep your model choice in configuration so a verified M3.1 API can replace M3 without changing your whole client. The M3 Pro status page follows the separate flagship; the H3.1 status page follows the other roadmap name.
FAQ
Is MiniMax M3.1 released?
The M3.1-Flash-Preview is live in MiniMax Code as of September 27, 2026. The complete M3.1 API release has not been documented or separately priced.
Can I call M3.1-Flash-Preview through APIMaster?
No public APIMaster route appeared in the September 27 check. The Code selector label is not a verified API model ID. Use M3's live card for an API available now.
Is this the roughly 3T M3 Pro?
No. The approximately 3T parameter and MSA 2.0 claims concern reports about a future M3 Pro, not the released Flash Preview. Neither rumored specification has been confirmed for the final product.
What does the check-in promotion cover?
MiniMax says the September 28–October 7 double daily free quota applies to M3.1-Flash-Preview, H3 and H3 Max. It is a MiniMax plan campaign, not a public API price guarantee.
Sources
- MiniMax Code — provider product and September 27 release announcement shown in MiniMax's official social channels.
- MiniMax M3 release notes — existing M3 product baseline; do not treat these as M3.1 specifications.
- APIMaster live M3 card — active-route count and current marketplace price (checked September 27, 2026).
Use MiniMax Code to try the preview. For a callable M3 API, register with APIMaster, create a key and test minimax-m3 through https://apimaster.ai/v1.