Skip to main content
AI Choice Engine

MiniMax

MiniMax M2.5

MiniMax M2.5 is a MiniMax model with 197K context, Not verified in / Not verified out per million tokens, last verified October 25, 2026. Open weights: yes. Status: Legacy; the current pick is MiniMax M3.1 Flash. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

MiniMax M2.5 is a legacy text-only downloadable checkpoint with 196,608 native positions under its specific MiniMax Model License.

Recorded profile verification October 25, 2026Individual provider fields may change

AI model specification details
SpecificationMiniMax M2.5
ProviderMiniMax
TierBalanced
Context window197K
Max outputNot verifiedUnverified
Input / 1M tokensNot verifiedUnverified
Output / 1M tokensNot verifiedUnverified
WeightsOpen
ParametersNot verifiedUnverified
Reasoning levelsNot verifiedUnverified
Modalitiestext
LicenseMiniMax Model License (M2.5)
ReleasedFebruary 12, 2026

Pricing tiers: Historical downloadable checkpoint: current first-party pricing and model overview no longer supply an M2.5 tariff. No current PAYG quote or endpoint availability is established here, and omission alone does not prove retirement. The current native config has 196,608 positions, distinct from the old hosted 204,800-token contract. Its specific MiniMax Model License governs weights, separately from serving terms.

Best for

  • Open-weight deployments
  • Text-only coding

Watch out

Native config is 196,608 positions, not the former hosted 204,800 context. Current first-party tariff and serving access are unverified. Redistribution requires the Model License, changed-file notices and attribution; prohibited-use terms apply. The card metadata says Modified-MIT, while the governing file is a specific MiniMax Model License. February 12 is announcement history; February 13 is the licence version date. Exact parameter counts and a separate native output cap remain unknown.

Source receipts

Catalog figures for this model were checked against the following sources.

Common questions

MiniMax M2.5

Answered from the verified figures on this page rather than general guidance.

What is MiniMax M2.5's context window?
MiniMax M2.5 accepts about 197K tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is MiniMax M2.5 best for?
MiniMax M2.5 is a balanced tier from MiniMax. It suits open-weight deployments, text-only coding. Native config is 196,608 positions, not the former hosted 204,800 context. Current first-party tariff and serving access are unverified. Redistribution requires the Model License, changed-file notices and attribution; prohibited-use terms apply. The card metadata says Modified-MIT, while the governing file is a specific MiniMax Model License. February 12 is announcement history; February 13 is the licence version date. Exact parameter counts and a separate native output cap remain unknown.
Can I self-host MiniMax M2.5?
MiniMax M2.5 publishes open weights, but self-hosting depends on the licence, hardware footprint, quantisation quality, and serving stack. A hosted API is often cheaper until you have measured throughput and concurrency on your own hardware.
Model comparison catalogBenchmark library