MiniMax M3 Pro: Release Status, Reported Specs, and API Alternatives
M3 Pro remains unreleased. Separate the reported ~3T and MSA 2.0 claims from MiniMax's M3.1 Flash Preview launch, and see a MiniMax M3 API with live pricing.
Published 2026-09-27
MiniMax M3 Pro has not been released as of September 27, 2026. An earlier earnings-call roadmap named M3.1, M3 Pro and H3.1. Reports around M3 Pro mention approximately 3T parameters and MSA 2.0, but these are forward-looking claims, not final public specifications. What launched on September 27 is M3.1-Flash-Preview inside MiniMax Code for fast everyday development. That preview is not M3 Pro, nor a full M3.1 public API announcement.
Need a working MiniMax text API? APIMaster currently lists minimax-m3 across 3 active routes, from $0.30 input / $1.20 output per 1M tokens versus the card's original $0.60 / $2.40, about 50% off (checked September 27, 2026). View the M3 discount; the live marketplace card is authoritative for today's price and route availability.
| Model ID | Original price (1M input / output) | APIMaster current price | Active routes | Discount | Model card |
|---|---|---|---|---|---|
| M3 Pro (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
Has MiniMax M3 Pro launched?
No. There is no published, callable M3 Pro API model ID, token price, context limit, maximum output, cache schedule or release date to put in production configuration. MiniMax's roadmap reference is evidence of a planned family, not of a launched endpoint. A community screenshot, checkpoint label or estimated parameter count cannot fill those operational gaps.
MiniMax's September 27 update concerns a different product: the lightweight M3.1-Flash-Preview in MiniMax Code. It is intended for routine software development, such as repairing bugs and implementing features. The Token Plan quota reset and September 28–October 7 check-in offer apply on MiniMax's own product surface; they are not a Pro launch or a Pro API price. Read the M3.1 Flash Preview guide for that release's access details.
What do the 3T and MSA 2.0 reports mean?
The community has discussed a roughly three-trillion-parameter M3 Pro and MSA 2.0 as part of a larger-model path. Neither figure should be treated as a published hardware requirement or performance guarantee. Parameter counting depends on whether a number refers to total or active parameters; an architecture label without an official technical report says even less about throughput and memory requirements.
The right comparison will be task-based. If M3 Pro appears, measure completed agent tasks, tool retries, edit accuracy, latency and total output tokens against M3 and the Flash Preview. Before then, any chart of model performance, context or API cost would be speculative.
| Question | M3 Pro status |
|---|---|
| Placement in MiniMax's pipeline | Named alongside M3.1 and H3.1 in an earlier earnings-call roadmap |
| Approximately 3T parameters | Reported expectation, not a final official specification |
| MSA 2.0 | Reported expectation, not publicly documented for a released endpoint |
| Public API model ID | Not announced |
| API price, cached input, context, output | Not announced |
| Release date | Not announced |
M3 Pro vs M3.1-Flash-Preview vs M3
M3 Pro is the unreleased larger tier under discussion. M3.1-Flash-Preview is the preview now usable in MiniMax Code. M3 (minimax-m3) is the existing text-model route on APIMaster that developers can test with an API key. None of these names should be silently substituted for another in a code sample.
If your immediate need is a fast development assistant, try the Flash Preview in MiniMax Code and log how often it fixes tests without broad edits. If you need a callable HTTP model today, use M3 and evaluate it on the same tasks. Wait for a provider model card and route health history before treating M3 Pro as a production dependency.
MiniMax API price you can actually use today
APIMaster's live minimax-m3 card has 3 active routes. The lowest lists $0.30 / $1.20 per 1M input / output tokens; the marketplace shows $0.60 / $2.40 as its original price. That is about 50% off for M3, not a prediction of M3 Pro's official price. Check the current card for route-specific rates, because availability, upstream terms and final billing can change.
To call M3, register, create a key, and use its real model ID:
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":"Suggest a focused fix for this failing test."}]}'
Do not replace minimax-m3 with a guessed minimax-m3-pro identifier. A published model card and a successful low-volume call should come first.
What to verify at launch
- Identity and availability: official name, exact API ID, release stage, regions and public route status.
- Economics: input, output and cache rates, context-based multipliers, free-plan eligibility and rate limits.
- Capability: published context and output limits, tools, multimodal support and benchmark methodology.
- Agent reliability: end-to-end task success, rollback rate, latency and token usage on your own codebase.
The H3.1 roadmap page tracks the other named future model without confusing it with the already available H3 video API.
FAQ
When is MiniMax M3 Pro coming out?
MiniMax has not announced a date for a public M3 Pro API. The roadmap reference does not commit to a particular day or interface.
Does M3 Pro have 3T parameters and MSA 2.0?
Those are community-reported expectations. No final MiniMax M3 Pro specification or independent measurement has established them as released-model facts.
Is M3.1-Flash-Preview M3 Pro?
No. The Flash Preview is a lighter MiniMax Code option released September 27; M3 Pro is a separate, unreleased roadmap model.
Is there an M3 Pro API price?
Not yet. APIMaster's $0.30 / $1.20 starting rate is for the existing minimax-m3 route and must not be presented as a Pro price.
Sources
- MiniMax — provider product and official September 27 Flash Preview announcement.
- MiniMax M3 release notes — currently available M3 baseline, not an M3 Pro specification.
- APIMaster M3 card — live routes and pricing checked September 27, 2026.
For today's API, register on APIMaster, create a key and call minimax-m3 at https://apimaster.ai/v1; keep the M3 Pro configuration pending official documentation.