Monthly cost, every model
Pick a workload and see what each flagship model would actually cost you per month. Bars are sorted cheapest → priciest. Numbers assume list prices without prompt caching.
Coding agent (5k reqs, 8k/2k tok)
Qwen: Qwen3.7 Flash$2.5DeepSeek: DeepSeek V4.1 Flash (batch)$3.5Anthropic: Claude Haiku 5.5 (batch)$4.5OpenAI: GPT-6 Luna Pro (batch)$4.5OpenAI: GPT-6 Luna (batch)$4.5DeepSeek: DeepSeek V4.1 Flash$5.7OpenAI: GPT-5 Nano$6.0Google: Gemini 2.5 Flash Lite$8.0Anthropic: Claude Haiku 5.5$9.0OpenAI: GPT-6 Luna Pro$9.0OpenAI: GPT-6 Luna$9.0Qwen: Qwen3.8 Omni Flash$10.7Qwen: Qwen3.8 Flash$10.7DeepSeek: DeepSeek V4.1 Flash 1017$12.0Meta: Llama 4 Maverick$14.0DeepSeek: DeepSeek V4 Flash Vision Exp$15.1Qwen: Qwen3.8 27B$15.1DeepSeek: DeepSeek V3.1$19.5OpenAI: GPT-5.6 Luna$20.0OpenAI: GPT-5.4 Nano$20.5OpenAI: GPT-5 Mini$30.0DeepSeek: DeepSeek V4 Flash 0731$30.8Google: Gemini 3.8 Flash (batch)$33.8Google: Gemini 3.7 Flash (batch)$33.8Google: Gemini 3.5 Flash Lite$37.0Google: Gemini 2.5 Flash$37.0DeepSeek: DeepSeek V4 Pro 0813$46.2Amazon: Nova Pro 1.0$64.0Google: Gemini 3.7 Flash$67.5Google: Gemini 3.6 Flash$67.5Google: Gemini 3.8 Flash$74.3OpenAI: GPT-5.4 Mini$75.0OpenAI: o4 Mini$88.0Google: Gemini 4 Argon (batch)$90.0OpenAI: GPT-6.1 Sol Pro (batch)$90.0OpenAI: GPT-6.1 Sol (batch)$90.0Anthropic: Claude Sonnet 5.5 (batch)$90.0OpenAI: GPT-6 Sol Pro (batch)$90.0OpenAI: GPT-6 Sol (batch)$90.0Mistral: Mistral Large 4$96.2Google: Nano Banana 2.1$135xAI: Grok 4.7$140Qwen: Qwen3.8 Max (0902)$140Qwen: Qwen3.8 2.4T A95B$140xAI: Grok 4.6$140Google: Gemini 3.5 Flash$150OpenAI: GPT-5$150Google: Gemini 2.5 Pro$150OpenAI: o3$160Anthropic: Claude Sonnet 5$180Google: Gemini 4 Argon$180OpenAI: GPT-6.1 Sol Pro$180OpenAI: GPT-6.1 Sol$180Anthropic: Claude Sonnet 5.5$180OpenAI: GPT-6 Sol Pro$180OpenAI: GPT-6 Sol$180Anthropic: Claude Opus 5.5 (batch)$180OpenAI: GPT-5.6 Terra Pro$200Anthropic: Claude Opus 5 (batch)$225OpenAI: GPT-5.4$250Anthropic: Claude Sonnet 4.5$270Qwen: Qwen3.8 Max Prime$280OpenAI: GPT-5.6 Sol$360Anthropic: Claude Opus 5.5$360Anthropic: Claude Opus 5$450OpenAI: GPT-6 Astra (batch)$450OpenAI: GPT-6 Astra Pro (batch)$450Anthropic: Claude Fable 5.1 (batch)$450OpenAI: GPT-5.5$500Anthropic: Claude Fable 5$900OpenAI: GPT-6 Astra$900OpenAI: GPT-6 Astra Pro$900Anthropic: Claude Fable 5.1$900Anthropic: Claude Opus 4.1$1,350OpenAI: GPT-5 Pro$1,800OpenAI: GPT-5.5 Pro$3,000
For this workload, Qwen: Qwen3.7 Flash costs $2.5/mo vs OpenAI: GPT-5.5 Pro at $3,000/mo — that's 1200× more expensive.
How we compute this
Monthly cost = requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000. Workload presets mirror real-world traffic shapes — see the scenario pages for full breakdowns, or the calculator to plug in your own numbers.