LLMCost

Chatbot LLM Cost

Customer-facing or internal chat assistants. Monthly estimates at 10,000 requests/mo, avg 800 input / 400 output tokens per request.

#ModelEst. monthly costvs cheapest
1Qwen: Qwen3.7 Flash$0.760—
2DeepSeek: DeepSeek V4.1 Flash (batch)$1.001.3×
3Anthropic: Claude Haiku 5.5 (batch)$1.401.8×
4OpenAI: GPT-6 Luna Pro (batch)$1.401.8×
5OpenAI: GPT-6 Luna (batch)$1.401.8×
6DeepSeek: DeepSeek V4.1 Flash$1.892.5×
7OpenAI: GPT-5 Nano$2.002.6×
8Google: Gemini 2.5 Flash Lite$2.403.2×
9Anthropic: Claude Haiku 5.5$2.803.7×
10OpenAI: GPT-6 Luna Pro$2.803.7×
11OpenAI: GPT-6 Luna$2.803.7×
12Qwen: Qwen3.8 Omni Flash$3.084.1×
13Qwen: Qwen3.8 Flash$3.084.1×
14DeepSeek: DeepSeek V4.1 Flash 1017$3.604.7×
15Meta: Llama 4 Maverick$4.115.4×
16DeepSeek: DeepSeek V4 Flash Vision Exp$4.315.7×
17Qwen: Qwen3.8 27B$5.727.5×
18DeepSeek: DeepSeek V3.1$5.807.6×
19OpenAI: GPT-5.6 Luna$6.408.4×
20OpenAI: GPT-5.4 Nano$6.608.7×
21DeepSeek: DeepSeek V4 Flash 0731$8.8011.6×
22OpenAI: GPT-5 Mini$10.0013.2×
23Google: Gemini 3.8 Flash (batch)$10.5013.8×
24Google: Gemini 3.7 Flash (batch)$10.5013.8×
25Google: Gemini 3.5 Flash Lite$12.4016.3×
26Google: Gemini 2.5 Flash$12.4016.3×
27DeepSeek: DeepSeek V4 Pro 0813$13.2017.4×
28Amazon: Nova Pro 1.0$19.2025.3×
29Google: Gemini 3.7 Flash$21.0027.6×
30Google: Gemini 3.6 Flash$21.0027.6×
31Google: Gemini 3.8 Flash$23.1030.4×
32OpenAI: GPT-5.4 Mini$24.0031.6×
33OpenAI: o4 Mini$26.4034.7×
34Mistral: Mistral Large 4$27.6036.3×
35Google: Gemini 4 Argon (batch)$28.0036.8×
36OpenAI: GPT-6.1 Sol Pro (batch)$28.0036.8×
37OpenAI: GPT-6.1 Sol (batch)$28.0036.8×
38Anthropic: Claude Sonnet 5.5 (batch)$28.0036.8×
39OpenAI: GPT-6 Sol Pro (batch)$28.0036.8×
40OpenAI: GPT-6 Sol (batch)$28.0036.8×
41xAI: Grok 4.7$40.0052.6×
42Qwen: Qwen3.8 Max (0902)$40.0052.6×
43Qwen: Qwen3.8 2.4T A95B$40.0052.6×
44xAI: Grok 4.6$40.0052.6×
45Google: Nano Banana 2.1$42.0055.3×
46Google: Gemini 3.5 Flash$48.0063.2×
47OpenAI: o3$48.0063.2×
48OpenAI: GPT-5$50.0065.8×
49Google: Gemini 2.5 Pro$50.0065.8×
50Anthropic: Claude Sonnet 5$56.0073.7×
51Google: Gemini 4 Argon$56.0073.7×
52OpenAI: GPT-6.1 Sol Pro$56.0073.7×
53OpenAI: GPT-6.1 Sol$56.0073.7×
54Anthropic: Claude Sonnet 5.5$56.0073.7×
55OpenAI: GPT-6 Sol Pro$56.0073.7×
56OpenAI: GPT-6 Sol$56.0073.7×
57Anthropic: Claude Opus 5.5 (batch)$56.0073.7×
58OpenAI: GPT-5.6 Terra Pro$64.0084.2×
59Anthropic: Claude Opus 5 (batch)$70.0092.1×
60OpenAI: GPT-5.4$80.00105.3×
61Qwen: Qwen3.8 Max Prime$80.00105.3×
62Anthropic: Claude Sonnet 4.5$84.00110.5×
63OpenAI: GPT-5.6 Sol$112.00147.4×
64Anthropic: Claude Opus 5.5$112.00147.4×
65Anthropic: Claude Opus 5$140.00184.2×
66OpenAI: GPT-6 Astra (batch)$140.00184.2×
67OpenAI: GPT-6 Astra Pro (batch)$140.00184.2×
68Anthropic: Claude Fable 5.1 (batch)$140.00184.2×
69OpenAI: GPT-5.5$160.00210.5×
70Anthropic: Claude Fable 5$280.00368.4×
71OpenAI: GPT-6 Astra$280.00368.4×
72OpenAI: GPT-6 Astra Pro$280.00368.4×
73Anthropic: Claude Fable 5.1$280.00368.4×
74Anthropic: Claude Opus 4.1$420.00552.6×
75OpenAI: GPT-5 Pro$600.00789.5×
76OpenAI: GPT-5.5 Pro$960.001263.2×

Cost-saving tips for chatbots

  • Cache your system prompt — repeated context is where caching saves 90%.
  • Route simple greetings to a mini/nano tier; escalate only on complex queries.
  • Cap max_output_tokens to prevent runaway responses.

Customize the numbers with the interactive calculator · All scenarios