getcheapaicredits

Live · refreshed hourly

The cheapest LLM APIs right now

Ranked by blended price per million tokens (3 input for 1 output), within each tier, among models released by the major labs in the last year. Cheaper isn't better at everything: compare within the tier you need.

Frontier (blended $5+ per 1M)

Cheapest first. Blended = (3 × input + output) / 4.

Mid-range ($1–5 per 1M)

ModelInput / 1MOutput / 1M
Z.ai: GLM 4.7$0.6$2.20
Z.ai: GLM 5.3$0.222$3.39
Qwen: Qwen3.6 27B$0.32$3.20
OpenAI: GPT Audio Mini$0.6$2.40
Qwen: Qwen3.8 27B$0.42$3
MoonshotAI: Kimi K2 Thinking$0.6$2.50
Google: Nano Banana 2 (Gemini 3.1 Flash Image)$0.5$3
Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)$0.5$3
Google: Gemini 3 Flash Preview$0.5$3
SpaceXAI: Grok Build 0.1$1$2

Cheapest first. Blended = (3 × input + output) / 4.

Small and cheap (under $1 per 1M)

ModelInput / 1MOutput / 1M
DeepSeek: DeepSeek V4 Flash 0423$0.042$0.084
Qwen: Qwen3.7 Flash$0.03$0.13
Mistral: Ministral 3 3B 2512$0.1$0.1
Qwen: Qwen3.5-9B$0.1$0.15
Qwen: Qwen3.5-Flash$0.065$0.26
OpenAI: gpt-oss-safeguard-20b$0.075$0.3
Google: Gemma 4 26B A4B $0.09$0.3
Z.ai: GLM 4.7 Flash$0.061$0.4
DeepSeek: DeepSeek V4.1 Flash$0.03$0.5
Mistral: Ministral 3 8B 2512$0.15$0.15

Cheapest first. Blended = (3 × input + output) / 4.

Cheaper still

The per-token price is the starting point. Batch APIs and caching cut it by up to 90% on repeated input, and where you buy credits changes what each dollar of usage costs. Full list: every model's price.