LLMCost

Customer Support Bot LLM Cost

High-volume ticket triage and first-line responses. Monthly estimates at 50,000 requests/mo, avg 1,200 input / 350 output tokens per request.

#ModelEst. monthly costvs cheapest
1Qwen: Qwen3.7 Flash$4.08—
2DeepSeek: DeepSeek V4.1 Flash (batch)$5.631.4×
3Anthropic: Claude Haiku 5.5 (batch)$7.381.8×
4OpenAI: GPT-6 Luna Pro (batch)$7.381.8×
5OpenAI: GPT-6 Luna (batch)$7.381.8×
6DeepSeek: DeepSeek V4.1 Flash$9.502.3×
7OpenAI: GPT-5 Nano$10.002.5×
8Google: Gemini 2.5 Flash Lite$13.003.2×
9Anthropic: Claude Haiku 5.5$14.753.6×
10OpenAI: GPT-6 Luna Pro$14.753.6×
11OpenAI: GPT-6 Luna$14.753.6×
12Qwen: Qwen3.8 Omni Flash$17.234.2×
13Qwen: Qwen3.8 Flash$17.234.2×
14DeepSeek: DeepSeek V4.1 Flash 1017$19.504.8×
15Meta: Llama 4 Maverick$22.675.6×
16DeepSeek: DeepSeek V4 Flash Vision Exp$24.256.0×
17Qwen: Qwen3.8 27B$26.026.4×
18DeepSeek: DeepSeek V3.1$31.637.8×
19OpenAI: GPT-5.6 Luna$33.008.1×
20OpenAI: GPT-5.4 Nano$33.888.3×
21DeepSeek: DeepSeek V4 Flash 0731$49.5012.1×
22OpenAI: GPT-5 Mini$50.0012.3×
23Google: Gemini 3.8 Flash (batch)$55.3113.6×
24Google: Gemini 3.7 Flash (batch)$55.3113.6×
25Google: Gemini 3.5 Flash Lite$61.7515.2×
26Google: Gemini 2.5 Flash$61.7515.2×
27DeepSeek: DeepSeek V4 Pro 0813$74.2518.2×
28Amazon: Nova Pro 1.0$104.0025.5×
29Google: Gemini 3.7 Flash$110.6327.1×
30Google: Gemini 3.6 Flash$110.6327.1×
31Google: Gemini 3.8 Flash$121.6929.9×
32OpenAI: GPT-5.4 Mini$123.7530.4×
33OpenAI: o4 Mini$143.0035.1×
34Google: Gemini 4 Argon (batch)$147.5036.2×
35OpenAI: GPT-6.1 Sol Pro (batch)$147.5036.2×
36OpenAI: GPT-6.1 Sol (batch)$147.5036.2×
37Anthropic: Claude Sonnet 5.5 (batch)$147.5036.2×
38OpenAI: GPT-6 Sol Pro (batch)$147.5036.2×
39OpenAI: GPT-6 Sol (batch)$147.5036.2×
40Mistral: Mistral Large 4$154.7538.0×
41Google: Nano Banana 2.1$221.2554.3×
42xAI: Grok 4.7$225.0055.2×
43Qwen: Qwen3.8 Max (0902)$225.0055.2×
44Qwen: Qwen3.8 2.4T A95B$225.0055.2×
45xAI: Grok 4.6$225.0055.2×
46Google: Gemini 3.5 Flash$247.5060.7×
47OpenAI: GPT-5$250.0061.3×
48Google: Gemini 2.5 Pro$250.0061.3×
49OpenAI: o3$260.0063.8×
50Anthropic: Claude Sonnet 5$295.0072.4×
51Google: Gemini 4 Argon$295.0072.4×
52OpenAI: GPT-6.1 Sol Pro$295.0072.4×
53OpenAI: GPT-6.1 Sol$295.0072.4×
54Anthropic: Claude Sonnet 5.5$295.0072.4×
55OpenAI: GPT-6 Sol Pro$295.0072.4×
56OpenAI: GPT-6 Sol$295.0072.4×
57Anthropic: Claude Opus 5.5 (batch)$295.0072.4×
58OpenAI: GPT-5.6 Terra Pro$330.0081.0×
59Anthropic: Claude Opus 5 (batch)$368.7590.5×
60OpenAI: GPT-5.4$412.50101.2×
61Anthropic: Claude Sonnet 4.5$442.50108.6×
62Qwen: Qwen3.8 Max Prime$450.00110.4×
63OpenAI: GPT-5.6 Sol$590.00144.8×
64Anthropic: Claude Opus 5.5$590.00144.8×
65Anthropic: Claude Opus 5$737.50181.0×
66OpenAI: GPT-6 Astra (batch)$737.50181.0×
67OpenAI: GPT-6 Astra Pro (batch)$737.50181.0×
68Anthropic: Claude Fable 5.1 (batch)$737.50181.0×
69OpenAI: GPT-5.5$825.00202.5×
70Anthropic: Claude Fable 5$1475.00362.0×
71OpenAI: GPT-6 Astra$1475.00362.0×
72OpenAI: GPT-6 Astra Pro$1475.00362.0×
73Anthropic: Claude Fable 5.1$1475.00362.0×
74Anthropic: Claude Opus 4.1$2212.50542.9×
75OpenAI: GPT-5 Pro$3000.00736.2×
76OpenAI: GPT-5.5 Pro$4950.001214.7×

Cost-saving tips for customer support bots

  • Cache your knowledge-base snippets — support queries share massive overlap.
  • Route escalations to a frontier model; let budget models handle FAQs.
  • Short outputs keep cost per ticket low — cap max_output_tokens.

Customize the numbers with the interactive calculator · All scenarios