US alternatives to MiniMax
8 non-Chinese options for buyers in the United States. See the MiniMax profile.
Compare MiniMax side by side with 3 of them →
OpenAI
United States
Alternative to Chinese LLMs
GPT / o-series models
- ✓ Available in US
- ✓ US-based vendor
- ✓ US data residency
Muse Spark Open
United States
Alternative to MiniMax
Meta's 816B open-weight model (2026), trained by Meta, not built on a Chinese base model
- 816B open weights
- ✓ Available in US
- ✓ US-based vendor
Reflection Beam
United States
Alternative to MiniMax
Reflection AI's 501B open-weight model (2026), trained from scratch
- 501B open weights, Apache 2.0
- ✓ Available in US
- ✓ US-based vendor
NVIDIA Nemotron
United States
Alternative to MiniMax
NVIDIA's open models, up to the 550B Nemotron 3 Ultra, pretrained by NVIDIA
- Up to 550B open weights
- ✓ Available in US
- ✓ US-based vendor
Arcee Trinity Large
United States
Alternative to MiniMax
Arcee's 400B open-weight model, trained from scratch in the US
- 400B open weights
- ✓ Available in US
- ✓ US-based vendor
gpt-oss
United States
Alternative to MiniMax
OpenAI's Apache-licensed open-weight models (120B and 20B)
- 120B and 20B open weights, Apache 2.0
- ✓ Available in US
- ✓ US-based vendor
Gemma
United States
Alternative to MiniMax
Google's Apache-licensed open models, up to Gemma 4 31B
- Up to 31B open weights, Apache 2.0
- ✓ Available in US
- ✓ US-based vendor
Olmo
United States
Alternative to MiniMax
Ai2's fully open models: weights, training data and code
- 32B; open weights, data and code
- ✓ Available in US
- ✓ US-based vendor