One agent session
1M input tokens and 100K output tokens, no caching.
- Input1M x $0.30 = $0.30
- Output0.1M x $1.20 = $0.12
$0.42per session
MiniMax - Open weights
MiniMax-M3 pricing and specs
MiniMax-M3 is an open-weights model from MiniMax that costs $0.30 per 1M input tokens and $1.20 per 1M output tokens on the first-party API, with a 1M-token context window.
Last verified from platform.minimax.io. All sources
- Input / 1M
- $0.30
- Output / 1M
- $1.20
- Context
- 1M tokens
- Weights
- Open
Pricing per 1M tokens
| Item | Price / 1M tokens |
|---|---|
| InputPrompts up to 512,000 tokens | $0.30 |
| OutputPrompts up to 512,000 tokens | $1.20 |
| Cache readPrompts up to 512,000 tokens | $0.06 |
| Long-context inputPrompts over 512,000 tokens | $0.60 |
| Long-context outputPrompts over 512,000 tokens | $2.40 |
| Long-context cache readPrompts over 512,000 tokens | $0.12 |
Note: Prices are the charged Standard-tier rates under MiniMax's 'Permanent 50% off' label; struck-through list prices are $0.60 input / $2.40 output / $0.12 cache read up to 512k input tokens and $1.20 / $4.80 / $0.24 above. The pricing page states the tier boundary only as '512k input tokens'; aboveTokens is recorded as 512000, but the page does not say whether 512k means 512,000 or 524,288 (the same docs write the max_tokens ceiling as '524288 (512K)'). Priority tier (service_tier=priority) is 1.5x Standard: $0.45 / $1.80 / $0.09 cache read up to 512k and $0.90 / $3.60 / $0.18 above. No cache write price is listed for M3. max output is the documented max_tokens ceiling and thinking tokens count toward it; thinking is off by default and enabled with thinking type adaptive. Accepts text, image and video input. The weights license allows commercial use with a 'Built with MiniMax M3' notice, and requires prior written authorization above $20M yearly revenue. The gateway lists a 512000 context for this id and its price varies by provider.
Specs
- Context window
- 1,000,000 tokens
- Max output
- 524,288 tokens
- Release date
- Provider
- MiniMax
- API model id
MiniMax-M3- Vercel AI Gateway id
minimax/minimax-m3- Weights
- Open weights
- License
- MiniMax Community License (Hugging Face license tag: other / minimax-community)
- Weights download
- huggingface.co
What MiniMax-M3 costs in practice
Worked from the standard list rates above, with the math shown so you can swap in your own token counts. These use the base rates, which apply while each request's prompt stays at or under 512,000 tokens; above that, MiniMax-M3 bills $0.60 input and $2.40 output for the whole request.
A chat app
10,000 requests a day at 2K input and 500 output tokens each.
- Input20M x $0.30 = $6.00
- Output5M x $1.20 = $6.00
$12.00per day
$360.00 per 30-day month
The same session at a 90% cache hit
90% of the input is read from the prompt cache. The one-time cache write premium is not included.
- Cached input0.9M x $0.06 = $0.054
- Uncached input0.1M x $0.30 = $0.03
- Output0.1M x $1.20 = $0.12
$0.204per session
Compare with
| Model | Input / 1M | Output / 1M |
|---|---|---|
| MiniMax-M3 (this page) | $0.30 | $1.20 |
| DeepSeek V4.1 FlashDeepSeek | $0.30 | $1.20 |
| GPT-5.6 LunaOpenAI | $0.20 | $1.20 |
| MiniMax-M2.5MiniMax | $0.30 | $1.20 |
| MiniMax-M2.7MiniMax | $0.30 | $1.20 |
| MiniMax-M2.5-highspeedMiniMax | $0.60 | $2.40 |
| MiniMax-M2.7-highspeedMiniMax | $0.60 | $2.40 |
Compare
Side-by-side pages with both models' prices, specs, cost examples and like-for-like benchmark rows.
- MiniMax-M3 vs Claude Haiku 5.5Open weights vs closed
- MiniMax-M3 vs GPT-6 LunaOpen weights vs closed
- MiniMax-M3 vs DeepSeek V4.1 FlashSimilar price, same tier
- MiniMax-M3 vs GLM-5.3-FlashSimilar price, same tier
- MiniMax-M3 vs Qwen3.8-FlashSimilar price, same tier
- MiniMax-M3 vs Gemini 3.5 Flash-LiteSimilar price, same tier
- MiniMax-M3 vs MiniMax-M2.7Same provider, successor
- MiniMax-M3 vs Mistral Small 4Similar price, same tier
Related reading
- MiniMax H3: An Omni-Modal Video Model With Native Audio, 2K Output, and Open Weights ComingMiniMax H3 turns text, image, video and audio input into 2K video with native stereo sound at 0.80 CNY per second, with open weights promised to follow.
- MiniMax M2.5 for Developers: The Anthropic-Compatible Budget Frontier ModelMiniMax M2.5 hits 80.2% on SWE-bench Verified and plugs into the Anthropic SDK with two environment variables. Here is what you need to know before switching.
FAQ
How much does MiniMax-M3 cost per million tokens?
MiniMax-M3 costs $0.30 per 1M input tokens and $1.20 per 1M output tokens on MiniMax's first-party API, per platform.minimax.io on October 17, 2026. Per 1M tokens, cache reads cost $0.06. Prompts over 512,000 tokens are billed at $0.60 input and $2.40 output for the whole request.
What is MiniMax-M3's context window?
MiniMax-M3 has a 1,000,000-token context window and returns up to 524,288 output tokens per response.
Are MiniMax-M3's weights open?
Yes. MiniMax-M3's weights are published at https://huggingface.co/MiniMaxAI/MiniMax-M3 under the MiniMax Community License (Hugging Face license tag: other / minimax-community).
When was MiniMax-M3 released?
MiniMax released MiniMax-M3 on June 1, 2026.
Sources
Last verified from platform.minimax.io.
- https://platform.minimax.io/docs/guides/pricing
- https://platform.minimax.io/docs/pricing/overview
- https://platform.minimax.io/docs/guides/text-generation
- https://platform.minimax.io/docs/api-reference/text-anthropic-api
- https://platform.minimax.io/docs/api-reference/text-chat-anthropic
- https://platform.minimax.io/docs/release-notes/models
- https://huggingface.co/MiniMaxAI/MiniMax-M3
- https://huggingface.co/MiniMaxAI/MiniMax-M3/raw/main/LICENSE
Newsletter
Get model price changes as they ship
One email per week. New model releases, price cuts and what they mean for your bill.