What does "native spec behavior" mean in MiniMax M2.5?

MiniMax M2.5 automatically produces a structured breakdown of functions, data structures, and UI components before writing code. This specification phase reduces implementation errors and improves coherence across multi-file outputs.

How does MiniMax M2.5 handle unfamiliar codebases?

It adapts more effectively than M2.1 and solves problems with fewer search rounds. This makes it better at navigating repositories it hasn't seen before.

What platforms does MiniMax M2.5 support for full-stack development?

Web, Android, iOS, Windows, and Mac. The model covers the full development lifecycle across all five platforms.

How does MiniMax M2.5 compare to M2.1 on speed?

MiniMax M2.5 completes tasks about 37% faster than M2.1 through optimized token efficiency in its reasoning process.

Is there a faster variant of MiniMax M2.5?

Yes. Select `minimax/minimax-m2.5-highspeed` where your provider exposes it. It targets high tokens-per-second for latency-sensitive applications.

Can MiniMax M2.5 be used in multi-agent pipelines?

Yes. Its native spec behavior and planning capabilities make it well-suited as a planner or orchestrator in multi-agent systems.

MiniMax

MiniMax M2.5

MiniMax M2.5 is a third-generation agentic model from MiniMax that handles full-stack development across Web, Android, iOS, Windows, and Mac platforms. It supports a context window of 1M tokens, a max output of 131K tokens, and completes tasks about 37% faster than M2.1. Your use is subject to MiniMax's Terms & Privacy Policies.

View API reference

Input and output price: Prices from: Input $0.30, Output $1.20, Per 1M tokens
24h uptime: Loading AI Gateway uptime

Get API key Read the docs

import { streamText } from 'ai'

const result = streamText({
  model: 'minimax/minimax-m2.5',
  prompt: 'Why is the sky blue?'
})

Read docs

Overview Playground Providers Uptime Status Throughput Latency API Similar About FAQ

Copy link to headingPlayground

Try out MiniMax M2.5 by MiniMax. Usage is billed to your team at API rates. Free users (those who haven't made a payment) get $5 of credits every 30 days.

MiniMax M2.5

Copy link to headingProviders

Route requests across multiple providers. Copy a provider slug to set your preference. Visit the docs for more info. Using a provider means you agree to their terms, listed under Legal.

Checking availability for your team

Provider

Context	Max Output	Latency	Throughput	Input	Output	Cache	Web Search	Capabilities	ZDR	No Training	Free Tier	Release Date

MiniMax

Legal:Terms•Privacy

205K

131K

0.8 s

85 tps

$0.30/M+1 more

$1.20/M+1 more

Read$0.03/M

Write$0.38/M

—

02/12/2026

Bedrock

Legal:Terms•Privacy

2.1 s

42 tps

$0.30/M

$1.20/M

—

02/12/2026

Getting started

Call MiniMax M2.5 through AI Gateway with the AI SDK generateText and streamText functions, or through the OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages APIs by changing the base URL. AI Gateway authenticates the request and routes it to an available provider.

Install the AI SDK (pnpm add ai dotenv), create an API key from the API Keys page, and set it as AI_GATEWAY_API_KEY in your environment. Full setup is covered in the text generation quickstart.

index.ts

import { generateText } from 'ai';
import 'dotenv/config';

async function main() {
  const result = await generateText({
    model: 'minimax/minimax-m2.5',
    prompt: 'Why is the sky blue?',
  });

  console.log(result.text);
}

main().catch(console.error);

Top-level parameters

The same MiniMax M2.5 request in each API format AI Gateway supports.

top-level-params.ts

import { generateText } from 'ai';
import 'dotenv/config';

async function main() {
  const result = await generateText({
    model: 'minimax/minimax-m2.5',
    system: 'You are a concise technical assistant.',
    prompt: 'Summarize the tradeoffs between static generation and SSR.',
    maxOutputTokens: 1024,
  });

  console.log(result.text);
}

main().catch(console.error);

Standard parameters like prompt, messages, temperature, and tools work as documented in the AI SDK docs. These are the parameters with model-specific behavior.

Parameter	Type	Required	Description
`model`	`string`	Yes	Model ID in the form `creator/model`, e.g. `minimax/minimax-m2.5`. AI Gateway routes the request to an available provider.
`maxOutputTokens`	`number`	No	Hard cap on generated tokens. MiniMax M2.5 supports up to `131,000` output tokens. Reasoning tokens count toward this limit.
`reasoning`	`'provider-default' \| 'none' \| 'minimal' \| 'low' \| 'medium' \| 'high' \| 'xhigh'`	No	Provider-agnostic reasoning effort, available in AI SDK 7 or later. Maps to the provider’s native reasoning configuration; reasoning settings under `providerOptions` take precedence when both are set. See the Reasoning section below.
`providerOptions`	`Record<string, JSONValue>`	No	AI Gateway routing options under `gateway`, plus any provider-native options under the provider’s own namespace — see the table below.

Input limits

Input	Formats	Sources	Max count	Max size	Limits
Text	—	—	—	—	Prompt and response share the 1M-token context window

Provider options

Set AI Gateway routing options under providerOptions.gateway. For provider-specific options, pass them under the provider’s namespace as documented by the AI SDK.

Learn more in the AI SDK provider docs.

provider-options.ts

import { generateText } from 'ai';
import 'dotenv/config';

async function main() {
  const result = await generateText({
    model: 'minimax/minimax-m2.5',
    prompt: 'Why is the sky blue?',
    providerOptions: {
      gateway: {
        only: ['minimax', 'bedrock'],
      },
    },
  });

  console.log(result.text);
}

main().catch(console.error);

These AI Gateway routing options apply to every model. Provider-specific options pass through under the provider’s own namespace (for example providerOptions.anthropic) exactly as documented by the AI SDK.

Parameter	Type	Required	Description
`providerOptions.gateway.only`	`string[]`	No	Restrict routing to these provider slugs. Requests fail over only within the listed providers.
`providerOptions.gateway.order`	`string[]`	No	Preferred provider order. Listed providers are tried first; unlisted providers remain available as fallbacks.
`providerOptions.gateway.sort`	`'cost' \| 'ttft' \| 'tps'`	No	Rank candidate providers by price, time to first token, or tokens per second instead of the default routing order.
`providerOptions.gateway.zeroDataRetention`	`boolean`	No	Route only to providers with a zero-data-retention policy for this model.

Routing across providers

AI Gateway serves the same model through multiple providers and fails over automatically. order expresses a preference while keeping every provider eligible; only is a hard allowlist — if none of the listed providers are available the request fails instead of falling back.

Options under a provider's own namespace (for example providerOptions.anthropic) are forwarded to that provider with the request. Providers ignore option namespaces that don't apply to them, so it is safe to set provider options alongside gateway routing options.

Reasoning

AI Gateway bridges reasoning across every API format. The AI SDK exposes a provider-agnostic top-level reasoning level (none, minimal, low, medium, high, or xhigh); the Chat Completions and Responses formats take the same effort under reasoning.effort; and the Anthropic Messages format uses a native thinking token budget. Whichever you send, the gateway maps it to the target model’s native configuration, converting between effort levels and token budgets as needed. Reasoning-related settings under providerOptions take full precedence over the top-level reasoning value and are never merged. Reasoning tokens typically count toward your output-token usage, though how they’re reported and billed varies by provider.

Learn more in the AI Gateway reasoning guide.

reasoning.ts

import { generateText } from 'ai';
import 'dotenv/config';

async function main() {
  const result = await generateText({
    model: 'minimax/minimax-m2.5',
    prompt: 'Explain the Monty Hall problem step by step.',
    reasoning: 'high',
  });

  console.log(result.text);
}

main().catch(console.error);

Tool calling

Expose tools the model can call. Define each tool’s inputs with a Zod schema.

tool-calling.ts

import { generateText, tool } from 'ai';
import { z } from 'zod';
import 'dotenv/config';

async function main() {
  const result = await generateText({
    model: 'minimax/minimax-m2.5',
    prompt: 'What is the weather in San Francisco?',
    tools: {
      getWeather: tool({
        description: 'Get the current weather for a location',
        inputSchema: z.object({ location: z.string() }),
        execute: async ({ location }) => ({ location, temperatureC: 18 }),
      }),
    },
  });

  console.log(result.text);
}

main().catch(console.error);

Copy link to headingMore models by MiniMax

Model

Context	Latency	Throughput	Input	Output	Cache	Web Search	Capabilities	Providers	ZDR	No Training	Free Tier	Release Date

minimax/minimax-m3

0.9 s

180 tps

$0.24/M+1 more

$0.96/M+1 more

Read$0.05/M

—

05/31/2026

minimax/minimax-m2.7

205K

1.2 s

69 tps

$0.21/M+1 more

$0.84/M+1 more

Read$0.04/M

Write$0.38/M

—

03/18/2026

minimax/minimax-m2.7-highspeed

205K

1.1 s

55 tps

$0.60/M

$2.40/M

Read$0.06/M

Write$0.38/M

—

03/18/2026

minimax/minimax-m2.5-highspeed

205K

1.0 s

66 tps

$0.60/M

$2.40/M

Read$0.03/M

Write$0.38/M

—

02/12/2026

minimax/minimax-m2.1

205K

0.4 s

132 tps

$0.30/M

$1.20/M

Read$0.03/M

Write$0.38/M

—

12/23/2025

minimax/minimax-m2

205K

0.8 s

86 tps

$0.30/M

$1.20/M

Read$0.03/M

Write$0.38/M

—

10/27/2025

Copy link to headingAbout MiniMax M2.5

Released on February 12, 2026, MiniMax M2.5 takes a different approach to software development: it plans before it builds. The model natively breaks down functions, data structures, and UI design into a specification before generating implementation code. This reduces downstream errors and produces more coherent multi-file outputs than models that write code directly.

MiniMax M2.5 scores 80.2% on SWE-Bench Verified and 51.3% on Multi-SWE-Bench, with improvement on multi-file software engineering tasks. The model completes tasks about 37% faster than M2.1 through optimized reasoning token efficiency, using fewer intermediate steps before reaching a solution.

MiniMax M2.5 spans the full development lifecycle: system design, implementation, and code review across Web, Android, iOS, Windows, and Mac platforms. It adapts more effectively to unfamiliar codebases than earlier generations, reducing ramp-up cost when you point the model at a new repository.

Copy link to headingWhat To Consider When Choosing a Provider

Configuration: MiniMax M2.5's native spec behavior produces structured plans that can feed downstream pipeline stages. This makes it a fit for multi-agent systems where one model plans and others execute.
Zero Data Retention: Zero Data Retention is available for this model. It is offered on a per-provider and model basis. See the documentation for details.
Authentication: AI Gateway authenticates requests using an API key or OIDC token. You do not need to manage provider credentials directly.

Copy link to headingWhen to Use MiniMax M2.5

Best for

Full-stack feature development: Work spanning multiple files and platforms
Unfamiliar codebase onboarding: An AI agent that requires exploration before coding
Multi-agent planner role: Architectures where a planning model feeds an execution model
End-to-end project delivery: Full workflows from system design through implementation and code review
Efficient convergence: Workloads that previously required many search rounds to converge on a solution

Consider alternatives when

Raw inference speed: Speed matters more than planning depth, so consider M2.5-highspeed
Simple single-file edits: Tasks that don't benefit from upfront planning
Multi-agent orchestration: You need the coordination features introduced in M2.7

Copy link to headingConclusion

MiniMax M2.5 shifts the series toward architecture-first development. Rather than generating code faster, it reasons about structure first. For teams building complex, multi-platform software with AI agents, that planning capability translates into fewer iterations and more coherent outputs.

Copy link to headingFrequently Asked Questions

What does "native spec behavior" mean in MiniMax M2.5?
MiniMax M2.5 automatically produces a structured breakdown of functions, data structures, and UI components before writing code. This specification phase reduces implementation errors and improves coherence across multi-file outputs.
How does MiniMax M2.5 handle unfamiliar codebases?
It adapts more effectively than M2.1 and solves problems with fewer search rounds. This makes it better at navigating repositories it hasn't seen before.
What platforms does MiniMax M2.5 support for full-stack development?
Web, Android, iOS, Windows, and Mac. The model covers the full development lifecycle across all five platforms.
How does MiniMax M2.5 compare to M2.1 on speed?
MiniMax M2.5 completes tasks about 37% faster than M2.1 through optimized token efficiency in its reasoning process.
What are MiniMax M2.5's SWE-Bench scores?
MiniMax M2.5 scores 80.2% on SWE-Bench Verified and 51.3% on Multi-SWE-Bench.
Is there a faster variant of MiniMax M2.5?
Yes. Select minimax/minimax-m2.5-highspeed where your provider exposes it. It targets high tokens-per-second for latency-sensitive applications.
Can MiniMax M2.5 be used in multi-agent pipelines?
Yes. Its native spec behavior and planning capabilities make it well-suited as a planner or orchestrator in multi-agent systems.

MiniMax M2.5

Copy link to headingPlayground

Copy link to headingProviders

Copy link to headingUptime24 hours

Copy link to headingThroughput24 hours

Copy link to headingLatency24 hours

Getting started

Top-level parameters

Input limits

Provider options

Routing across providers

Reasoning

Tool calling

Copy link to headingMore models by MiniMax

Copy link to headingAbout MiniMax M2.5

Copy link to headingWhat To Consider When Choosing a Provider

Copy link to headingWhen to Use MiniMax M2.5

Best for

Consider alternatives when

Copy link to headingConclusion

Copy link to headingFrequently Asked Questions