LLMCost

Coding Agent LLM Cost

Cursor/Claude Code style dev assistants with large contexts. Monthly estimates at 5,000 requests/mo, avg 8,000 input / 2,000 output tokens per request.

#ModelEst. monthly costvs cheapest
1Qwen: Qwen3.7 Flash$2.50
2DeepSeek: DeepSeek V4 Flash 0731$5.002.0×
3OpenAI: GPT-5 Nano$6.002.4×
4Google: Gemini 2.5 Flash Lite$8.003.2×
5OpenAI: GPT-5.6 Luna Pro (batch)$10.004.0×
6OpenAI: GPT-5.6 Luna (batch)$10.004.0×
7DeepSeek: DeepSeek V4 Flash Vision Exp$15.406.2×
8Meta: Llama 4 Maverick$16.006.4×
9Google: Gemini 3.7 Flash (batch)$16.886.8×
10Google: Gemini 3.5 Flash Lite (batch)$18.507.4×
11OpenAI: GPT-5.6 Luna$20.008.0×
12OpenAI: GPT-5.6 Luna Pro$20.008.0×
13OpenAI: GPT-5.4 Nano$20.508.2×
14Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)$25.0010.0×
15Qwen: Qwen3.7 Plus$25.6010.2×
16OpenAI: GPT-5 Mini$30.0012.0×
17Google: Gemini 3.7 Flash$33.7513.5×
18Google: Gemini 3.6 Flash (batch)$33.7513.5×
19Qwen: Qwen3 235B A22B$36.4014.6×
20Google: Gemini 3.5 Flash Lite$37.0014.8×
21Google: Gemini 2.5 Flash$37.0014.8×
22DeepSeek: DeepSeek V3.1$38.5015.4×
23Qwen: Qwen3.8 27B$46.0018.4×
24Google: Nano Banana 2 (Gemini 3.1 Flash Image)$50.0020.0×
25DeepSeek: R1$53.0021.2×
26Amazon: Nova Pro 1.0$64.0025.6×
27Google: Gemini 3.6 Flash$67.5027.0×
28OpenAI: GPT-5.4 Mini$75.0030.0×
29DeepSeek: DeepSeek V4 Pro 0813$78.5431.4×
30OpenAI: o4 Mini$88.0035.2×
31OpenAI: GPT-5.6 Sol Pro (batch)$90.0036.0×
32OpenAI: GPT-5.6 Sol (batch)$90.0036.0×
33Anthropic: Claude Sonnet 5 (batch)$90.0036.0×
34OpenAI: GPT-5.6 Terra Pro (batch)$100.0040.0×
35OpenAI: GPT-5.6 Terra (batch)$100.0040.0×
36Qwen: Qwen3.8 2.4T A95B$140.0056.0×
37SpaceXAI: Grok 4.6$140.0056.0×
38Qwen: Qwen3.8 Max$140.0056.0×
39SpaceXAI: Grok 4.5$140.0056.0×
40Google: Gemini 3.5 Flash$150.0060.0×
41OpenAI: GPT-5$150.0060.0×
42Google: Gemini 2.5 Pro$150.0060.0×
43OpenAI: o3$160.0064.0×
44OpenAI: GPT-5.6 Sol$180.0072.0×
45Anthropic: Claude Sonnet 5$180.0072.0×
46OpenAI: GPT-5.6 Sol Pro$180.0072.0×
47OpenAI: GPT-5.6 Terra Pro$200.0080.0×
48OpenAI: GPT-5.6 Terra$200.0080.0×
49Google: Nano Banana Pro (Gemini 3 Pro Image)$200.0080.0×
50Claude Opus 5 (batch)$225.0090.0×
51Anthropic: Claude Opus 4.8 (batch)$225.0090.0×
52OpenAI: GPT-5.4$250.00100.0×
53Anthropic: Claude Sonnet 4.5$270.00108.0×
54Claude Opus 5$450.00180.0×
55Anthropic: Claude Fable 5 (batch)$450.00180.0×
56Anthropic: Claude Opus 4.8$450.00180.0×
57OpenAI: GPT-5.5$500.00200.0×
58Anthropic: Claude Fable 5$900.00360.0×
59Claude Opus 5 (Fast)$900.00360.0×
60Anthropic: Claude Opus 4.8 (Fast)$900.00360.0×
61Anthropic: Claude Opus 4.1$1350.00540.0×
62OpenAI: GPT-5 Pro$1800.00720.0×
63OpenAI: GPT-5.5 Pro$3000.001200.0×

Cost-saving tips for coding agents

  • Input dominates cost here (8K in vs 2K out) — prompt caching is your biggest lever.
  • Use a frontier model only for planning steps; route file edits to mid-tier.
  • Trim context: send relevant files, not the whole repo.

Customize the numbers with the interactive calculator · All scenarios