Cheapest LLMs by Price per Token
The cheapest LLMs are the large language models with the lowest API cost per token. Pricing is quoted per million (1M) tokens, with output tokens usually costing more than input. The table below ranks models from lowest to highest output price — a good starting point for high-volume, cost-sensitive workloads where you can trade some quality for price.
Ranked list (top 25)
| # | Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|---|
| 1 | Ambient Small AmbientOpen weights | $0.00 | $0.00 | 101K |
| 2 | qwen3.8-27b-4q AmbientOpen weights | $0.00 | $0.00 | 101K |
| 3 | KB Whisper evrocOpen weights | $0.00 | $0.00 | 448 |
| 4 | Whisper 3 Large evrocOpen weights | $0.00 | $0.00 | 448 |
| 5 | Llama 3.1 Nemotron Safety Guard VultrOpen weights | $0.01 | $0.01 | 8K |
| 6 | Meta Llama Prompt Guard 2 22M Helicone | $0.01 | $0.01 | 512 |
| 7 | Meta Llama Prompt Guard 2 86M Helicone | $0.01 | $0.01 | 512 |
| 8 | Qwen3 Embedding 0.6B NEAR AI CloudOpen weights | $0.01 | $0.01 | 33K |
| 9 | Qwen3 Reranker 0.6B NEAR AI CloudOpen weights | $0.01 | $0.01 | 41K |
| 10 | E5 Mistral 7B STACKITOpen weights | $0.02 | $0.02 | 4K |
| 11 | PaddleOCR VL Novita AIOpen weights | $0.02 | $0.02 | 16K |
| 12 | DeepSeek-OCR NovitaAIOpen weights | $0.03 | $0.03 | 8K |
| 13 | Google Gemma 2 Helicone | $0.01 | $0.03 | 8K |
| 14 | inclusionAI: Ling-2.6-flash inclusionAI | $0.01 | $0.03 | 262K |
| 15 | Llama 3.1 8B (decentralized) NanoGPTOpen weights | $0.02 | $0.03 | 128K |
| 16 | Llama Prompt Guard 2 22M GroqOpen weights | $0.03 | $0.03 | 512 |
| 17 | Llama-Guard-3-8B (Cloudflare) Eden AIOpen weights | $0.48 | $0.03 | 131K |
| 18 | Meta Llama 3.1 8B Instruct Turbo Helicone | $0.02 | $0.03 | 128K |
| 19 | Mistral: Mistral Nemo Mistral | $0.03 | $0.03 | 131K |
| 20 | Qwen3 4B NovitaAIOpen weights | $0.03 | $0.03 | 128K |
| 21 | Qwen3 4B FP8 LLM GatewayOpen weights | $0.03 | $0.03 | 128K |
| 22 | Qwen3-ASR Flash Alibaba | $0.04 | $0.04 | 53K |
| 23 | Ministral 3B (latest) MistralOpen weights | $0.04 | $0.04 | 131K |
| 24 | Prompt Guard 2 86M GroqOpen weights | $0.04 | $0.04 | 512 |
| 25 | Voxtral Mini 3B 2507 (Amazon Bedrock, US) Eden AIOpen weights | $0.04 | $0.04 | 33K |
Prices are per 1M tokens (USD); confirm with the provider. Updated regularly.
More LLM rankings
Frequently asked questions
What is the cheapest LLM?
Ranked by output price per 1M tokens, the cheapest LLM currently listed is Ambient Small. Pricing changes often, so check the full ranking above for the latest order.
Is the cheapest LLM good enough?
Cheapest is rarely best for every task. Low-cost models work well for classification, extraction, and high-volume summarization, but frontier reasoning tasks may justify a pricier model. Weigh price against benchmark quality and context window.
Why is output more expensive than input?
Generating tokens (output) is more compute-intensive than reading them (input), so providers price output higher. For cost estimates, weight expected output volume accordingly.