MiniMax M2 for OpenClaw: Pricing, Setup, and What It's Good At

Current as of March 2026. MiniMax M2 is a reasoning model from China’s MiniMax lab. The pricing is blunt: $0.30 input, $1.20 output, 200K context. It’s not trying to compete with Claude on quality — it’s trying to compete on cost-per-token for tasks where you need a long context and a reasoning pass.

Specs

Provider	MiniMax
Input cost	$0.30 / M tokens
Output cost	$1.20 / M tokens
Context window	200K tokens
Max output	8K tokens
Parameters	N/A
Features	function_calling, reasoning

What it’s good at

Price-to-Performance Ratio

$0.30/M input for a model with reasoning is genuinely cheap. You’re getting something that can handle multi-step logic without paying GPT-4o rates.

Context Handling

200K tokens and the retrieval quality holds up reasonably well across the window. I haven’t hit the severe needle-in-haystack degradation that plagues some of the cheaper models at this context size.

Where it falls short

Inconsistent Latency

Time-to-first-token varies more than I’d like, especially through aggregator routing. Don’t put this on a path where the user is waiting.

Dry Prose

The output is functional, not elegant. If you’re generating text that a human will read, expect to edit it. Claude this is not.

Best use cases with OpenClaw

• Large-Scale Document Analysis — 200K context + cheap input = a reasonable way to summarize entire codebases or legal archives without a RAG pipeline.
• Structured Data Extraction — The function calling works reliably, and at $1.20 output you can run this against a lot of records before costs get painful.

Not ideal for

• Creative Copywriting — The model defaults to repetitive sentence structures. Heavy editing required.
• Low-Latency Chatbots — Too much variance in TTFT for anything user-facing that needs to feel fast.

Run it through Haimaker

Skip juggling API keys. One Haimaker key gives you access to every model on the platform. Tell OpenClaw:

Add Haimaker as a custom provider to my OpenClaw config. Use these details:

- Provider name: haimaker
- Base URL: https://api.haimaker.ai/v1
- API key: [PASTE YOUR HAIMAKER API KEY HERE]
- API type: openai-completions

Add the auto-router model:
- haimaker/auto (reasoning: false, context: 128000, max tokens: 32000)

Create an alias "auto" for easy switching. Apply the config when done.

Or skip model selection entirely — Haimaker’s auto-router picks the best model for each task so you don’t have to.

OpenClaw setup

Point your OpenClaw base URL to https://api.haimaker.ai/v1 and use your Haimaker API key; the model follows standard OpenAI-compatible schemas for function calling.

{
  "models": {
    "mode": "merge",
    "providers": {
      "minimax": {
        "baseUrl": "https://api.haimaker.ai/v1",
        "apiKey": "YOUR-MINIMAX-API-KEY",
        "api": "openai-completions",
        "models": [
          {
            "id": "MiniMax-M2",
            "name": "MiniMax M2",
            "cost": {
              "input": 0.3,
              "output": 1.2
            },
            "contextWindow": 200000,
            "maxTokens": 8192
          }
        ]
      }
    }
  }
}

How it compares

• vs GPT-4o-mini — 4o-mini is slightly cheaper on input ($0.15/M vs $0.30/M) and has stronger reasoning, but its context window caps at 128K.
• vs DeepSeek-V3 — DeepSeek wins on coding. M2 tends to be steadier for general instruction following in longer agent loops.

Bottom line

M2 makes sense when you need a reasoning pass over a large document and you don’t want to pay GPT-4o prices to do it. Know what you’re trading: cost efficiency in, not quality out.

TRY MINIMAX M2 ON HAIMAKER

---

For setup instructions, see our API key guide. For all available models, see the complete models guide.