// MiniMax
MiniMax-M3
A MiniMax model. 10 prices across 10 channels side by side, each with its source and check date – plus specs, licence, speed and benchmarks.
- Official in/out per 1M
- ¥2.1 / ¥8.4
- Lowest (GMICloud)
- $0.24 / $0.96
- Context
- 1.05M
- AA index
- 29.2
Prices checked 2026-10-22Official sources linked on every rowNo estimatesCheapest LLM APIs →
01Specs at a glance
Official values first; gaps filled from OpenRouter’s API where the vendor page is silent.
| Vendor | MiniMax |
|---|---|
| Context window | 1.05M (1,048,576 tokens) |
| Max output | 512K |
| Input | Text · Image · Video |
| Output | Text |
| Tool calling / JSON | Yes / Yes |
| Weights & licence | Open weights · minimax-community (custom) · Hugging Face |
| Parameters | 427B (from weight files) |
| Released | May 31, 2026 |
| Model IDs | MiniMax-M3 · minimax/minimax-m3 · minimax-m3 · MiniMax-M3@bailian |
02Prices on 10 channels
Per 1M tokens, in the currency each source publishes (1 USD = 6.7 CNY for the blended column). Sorted: official first, then cheapest.
| Channel | Source | Input | Output | Cache hit | Blended $ | Context | Notes |
|---|---|---|---|---|---|---|---|
| Official Vendor’s China site, billed in CNY | Official 10-20 | ¥2.1 | ¥8.4 | ¥0.42 | 0.549 | 1.05M | prompt>512K: 4.2/16.8 Priority tier 1.5x (≤512K 3.15/12.60/0.63; >512K 6.30/25.20/1.26). Since 2026-10-20 the page shows these prices directly (the old “permanent 50% off” list price 4.2/16.8 is no longer shown). Intl: $0.30/$1.20/$0.06 (≤512K), $0.60/$2.40/$0.12 (>512K) |
| GMICloud · fp8 Via OpenRouter / provider listed there | OpenRouter 10-22 | $0.24 | $0.96 | $0.048 | 0.42 | 1.05M | GMICloud price as listed on OpenRouter (pass-through; quantization=fp8; discount field=0.2); not checked against GMICloud's own page |
| DeepInfra · fp8 Via OpenRouter / provider listed there | OpenRouter 10-22 | $0.28 | $1.1 | $0.056 | 0.485 | 524K | DeepInfra price as listed on OpenRouter (pass-through; quantization=fp8; discount field=0); not checked against DeepInfra's own page |
| Together Inference platform (own price page) | Platform 10-10 | $0.3 | $1.2 | $0.06 | 0.525 | 1.05M | Together AI own pricing page |
| Fireworks Inference platform (own price page) | Platform 10-10 | $0.3 | $1.2 | $0.06 | 0.525 | 1.05M | Batch −50%Fireworks AI own pricing page; batch -50% |
| Parasail · fp8 Via OpenRouter / provider listed there | OpenRouter 10-22 | $0.3 | $1.2 | $0.06 | 0.525 | 1.05M | Parasail price as listed on OpenRouter (pass-through; quantization=fp8; discount field=0); not checked against Parasail's own page |
| AtlasCloud · fp8 Via OpenRouter / provider listed there | OpenRouter 10-22 | $0.3 | $1.2 | $0.06 | 0.525 | 524K | AtlasCloud price as listed on OpenRouter (pass-through; quantization=fp8; discount field=0); not checked against AtlasCloud's own page |
| Novita · fp8 Via OpenRouter / provider listed there | OpenRouter 10-22 | $0.3 | $1.2 | $0.06 | 0.525 | 1M | Novita price as listed on OpenRouter (pass-through; quantization=fp8; discount field=0); not checked against Novita's own page |
| SambaNova · unknown Via OpenRouter / provider listed there | OpenRouter 10-22 | $0.6 | $2.4 | $0.06 | 1.05 | 1.05M | SambaNova price as listed on OpenRouter (pass-through; quantization=unknown; discount field=0); not checked against SambaNova's own page |
| 阿里云百炼 Inference platform (own price page) | Platform 10-10 | ¥4.2 | ¥16.8 | — | 1.1 | — | Third-party model hosted on 阿里云百炼 (Beijing) |
“Official” = vendor pricing page; “Platform” = the platform’s own pricing page; “OpenRouter” = price listed on OpenRouter, not checked on the platform’s site. Click a source tag to open the original page.
03Where you can use it
Derived only from the price sources above. Country restrictions are set by each vendor – check their terms before you sign up.
- Vendor’s China site, billed in CNY
Official - Via OpenRouter / provider listed there
GMICloud · DeepInfra · Parasail · AtlasCloud · Novita · SambaNova - Inference platform (own price page)
Together · Fireworks · 阿里云百炼 - Self-host the open weights
GPU / VRAM sizing and build plans→ - Free option
MiniMax-M3→
04Performance
All results →Speed and AA index from Artificial Analysis public leaderboards (not our own tests); benchmark scores from our benchmark library, best result per benchmark.
- AA indexMiniMax-M3
- 29.2
- Output speed (median)
- 87 tok/s
- Time to first token
- 1.41s
- AA index per $ (blended)
- 53.27
| Benchmark | Best score | Source | Date |
|---|---|---|---|
| LMArena Text | 1441 | official | 2026-10-08 |
| LMArena WebDev | 1482 | official | 2026-10-08 |
| LMArena Vision | 1237 | official | 2026-10-02 |
| GPQA Diamond | 81.3% | independent | 2026-08-10 |
| AIME (OTIS Mock AIME 2024–2025) | 26.7% | independent | 2026-08-10 |
| OSWorld 2.0 | 4.6% | official | 2026-06-01 |
05Price changes we recorded
All price changes →From our daily re-checks of the sources above.
No price change recorded since we started tracking (2026-10-10).
06More from MiniMax
FAQ
How much does the MiniMax-M3 API cost?
On the official pricing page: ¥2.1 / ¥8.4 per 1M input / output tokens, cached input ¥0.42 (checked 2026-10-20). The lowest price we found is $0.24 / $0.96 on GMICloud.
What is the context window of MiniMax-M3?
1.05M tokens of context, up to 512K output tokens. Platforms may cap it lower – see the table.
Is MiniMax-M3 open source?
Yes, the weights are published on Hugging Face (MiniMaxAI/Minimax-M3). Licence: minimax-community (custom).