🔄 Same Model, Different Prices — Find the Cheapest Host

The same open-source model varies widely in price across platforms — compare them all at once.

Llama 3.3 70B Instruct

70BOpen Source (Llama 3.3)Save up to 31%

Meta's latest open-source large model — great for chat and coding.

Platform	Input /1M	Output /1M	Context	Speed	Savings	Go
GroqCheapest llama-3.3-70b-versatile	$0.59	$0.79	128K	Ultra Fast	Save 31%	Go
Together AI meta-llama/Llama-3.3-70B-Instruct-Turbo	$0.88	$0.88	131K	—	Save 12%	Go
Fireworks AI accounts/fireworks/models/llama-v3p3-70b-instruct	$0.90	$0.90	131K	—	Save 10%	Go
Microsoft Llama-3.3-70B-Instruct	$0.90	$0.90	128K	—	Save 10%	Go
AWS Bedrock meta.llama3-3-70b-instruct-v1	$0.99	$0.99	128K	—	Save 1%	Go
Perplexity llama-3.3-70b-instruct	$1.00	$1.00	131K	—	Baseline	Go

Qwen 2.5 72B Instruct

72BOpen Source (Apache 2.0)Save up to 31%

Alibaba's most powerful model in the Qwen series.

Platform	Input /1M	Output /1M	Context	Speed	Savings	Go
Fireworks AICheapest accounts/fireworks/models/qwen2p5-72b-instruct	$0.90	$0.90	131K	—	Save 31%	Go
Together AI Qwen/Qwen2.5-72B-Instruct-Turbo	$1.20	$1.20	131K	—	Save 8%	Go
AWS Bedrock qwen2.5-72b-instruct	$1.30	$1.30	128K	—	Baseline	Go

Mistral Large

123BCommercial

Mistral's flagship model, available across multiple platforms.

Platform	Input /1M	Output /1M	Context	Speed	Savings	Go
MistralCheapest mistral-large-latest	$2.00	$6.00	128K	—	Baseline	Go
AWS Bedrock mistral.mistral-large-2407-v1	$2.00	$6.00	128K	—	Baseline	Go
Azure mistral-large-latest	$2.00	$6.00	128K	—	Baseline	Go
Microsoft Mistral-Large	$2.00	$6.00	128K	—	Baseline	Go

DeepSeek V4

MoEOpen SourceSave up to 77%

DeepSeek's latest-generation model with exceptional value.

Platform	Input /1M	Output /1M	Context	Speed	Savings	Go
DeepSeekCheapest deepseek-chat	$0.14	$0.28	131K	Fast	Save 77%	Go
Together AI deepseek-ai/DeepSeek-V3	$0.90	$0.90	131K	—	Baseline	Go
Fireworks AI accounts/fireworks/models/deepseek-v3	$0.90	$0.90	131K	—	Baseline	Go

Gemma 2 27B

27BOpen Source (Gemma)Save up to 75%

Google's open-source model — lightweight and efficient.

Platform	Input /1M	Output /1M	Context	Speed	Savings	Go
GroqCheapest gemma2-9b-it	$0.20	$0.20	8K	Ultra Fast	Save 75%	Go
Together AI google/gemma-2-27b-it	$0.80	$0.80	8K	—	Baseline	Go
Fireworks AI accounts/fireworks/models/gemma2-27b-it	$0.80	$0.80	8K	—	Baseline	Go

Pick the right host and open-source model costs can differ 2-5×!

The same Llama, Qwen, or DeepSeek model can cost far less just by switching hosts. Before you commit, compare prices and speeds across platforms to spend your budget wisely.