Monthly generation cost of a retrieval-augmented (RAG) Q&A app with 20,000 questions, across every major model.
Bottom line: cheapest is Meta: Llama 3.2 1B Instruct at about $2.18/month; the median model costs $49.55/month; the most expensive, Anthropic: Claude Fable 5, costs $678.30/month. That is a 311× spread for the same workload.
Scenario: RAG app: 20,000 questions/month, 5 retrieved chunks × 300 words + 50-word question in / 200 words out (generation only; embedding and vector-DB costs excluded).
| Vendor | Model | In $/1M | Out $/1M | Est. monthly cost | Source / check |
|---|---|---|---|---|---|
| Meta (Llama) | Meta: Llama 3.2 1B Instruct | $0.027 | $0.201 | $2.18 | source 2026-10-16 ✔ matches |
| Meta (Llama) | Meta: Llama 3.1 8B Instruct | $0.05 | $0.08 | $2.49 | source 2026-10-16 ✔ matches |
| Meta (Llama) | Meta: Llama 3.2 3B Instruct | $0.05 | $0.33 | $3.82 | source 2026-10-16 ✔ matches |
| DeepSeek | DeepSeek: DeepSeek V4 Flash 0731 | $0.0111 | $0.64 | $3.86 | source 2026-10-16 ℹ vs market median ($0.14/$0.4785) |
| DeepSeek | DeepSeek: DeepSeek V4.1 Flash | $0.018 | $0.684 | $4.38 | source 2026-10-16 ⚠ differs from lab price ($0.15/$0.6) |
| Mistral | Mistral: Ministral 3 3B 2512 | $0.1 | $0.1 | $4.66 | source 2026-10-16 ✔ matches |
| DeepSeek | DeepSeek: DeepSeek V4 Flash 0423 | $0.091 | $0.182 | $4.72 | source 2026-10-16 ℹ vs market median ($0.14/$0.28) |
| Anthropic | Anthropic: Claude Haiku 5.5 | $0.1 | $0.5 | $6.78 | source 2026-10-16 ✔ matches |
| OpenAI | OpenAI: GPT-6 Luna Pro | $0.1 | $0.5 | $6.78 | source 2026-10-16 ✔ matches |
| OpenAI | OpenAI: GPT-6 Luna | $0.1 | $0.5 | $6.78 | source 2026-10-16 ✔ matches |
| Mistral | Mistral: Ministral 3 8B 2512 | $0.15 | $0.15 | $6.98 | source 2026-10-16 ✔ matches |
| Meta (Llama) | Meta: Llama Guard 4 12B | $0.18 | $0.18 | $8.38 | source 2026-10-16 ✔ matches |
| Mistral | Mistral: Ministral 3 14B 2512 | $0.2 | $0.2 | $9.31 | source 2026-10-16 ✔ matches |
| Mistral | Mistral: Mistral Small 4 | $0.15 | $0.6 | $9.38 | source 2026-10-16 ✔ matches |
| Meta (Llama) | Meta: Llama 4 Scout | $0.18 | $0.59 | $10.56 | source 2026-10-16 ℹ vs market median ($0.1/$0.3139) |
| Meta (Llama) | Meta: Llama 4 Maverick | $0.1875 | $0.6525 | $11.20 | source 2026-10-16 ℹ vs market median ($0.2546/$0.85) |
| DeepSeek | DeepSeek: DeepSeek V4 Flash Vision Exp | $0.2156 | $0.6468 | $12.33 | source 2026-10-16 ℹ vs market median ($0.3321/$0.9963) |
| DeepSeek | DeepSeek: DeepSeek V3.2 Exp | $0.27 | $0.41 | $13.31 | source 2026-10-16 ⚠ differs from lab price ($0.28/$0.42) |
| DeepSeek | DeepSeek: DeepSeek V3.2 | $0.259 | $0.8 | $14.93 | source 2026-10-16 ⚠ differs from lab price ($0.28/$0.42) |
| Google: Gemini 3.1 Flash Lite | $0.25 | $1.5 | $18.29 | source 2026-10-16 ✔ matches | |
| Meta (Llama) | Meta: Llama 3.1 70B Instruct | $0.4 | $0.4 | $18.62 | source 2026-10-16 ℹ vs market median ($0.4/$0.56) |
| Google: Gemini 3.5 Flash Lite | $0.3 | $2.5 | $25.67 | source 2026-10-16 ✔ matches | |
| Mistral | Mistral: Devstral 2 2512 | $0.4 | $2.0 | $27.13 | source 2026-10-16 ✔ matches |
| Meta (Llama) | Meta: Llama 3.3 70B Instruct | $0.59 | $0.79 | $28.53 | source 2026-10-16 ℹ vs market median ($0.5683/$0.72) |
| Mistral | Mistral: Mistral Large 3 2512 | $0.5 | $1.5 | $28.59 | source 2026-10-16 ✔ matches |
| DeepSeek | DeepSeek: DeepSeek V4 Pro 0813 | $0.66 | $1.98 | $37.75 | source 2026-10-16 ✔ matches |
| Mistral | Mistral: Mistral Large 4 | $0.68 | $2.09 | $39.16 | source 2026-10-16 ✔ matches |
| DeepSeek | DeepSeek: DeepSeek V4 Pro 0423 | $0.9553 | $1.9105 | $49.55 | source 2026-10-16 ⚠ differs from lab price ($0.66/$1.98) |
| Google: Gemini 3.7 Flash | $0.75 | $3.75 | $50.87 | source 2026-10-16 ✔ matches | |
| Google: Gemini 3.6 Flash | $0.75 | $3.75 | $50.87 | source 2026-10-16 ✔ matches | |
| xAI | SpaceXAI: Grok Build 0.1 | $1.0 | $2.0 | $51.87 | source 2026-10-16 ✔ matches |
| Google: Gemini 3.8 Flash | $0.825 | $4.125 | $55.96 | source 2026-10-16 ⚠ differs from lab price ($0.75/$3.75) | |
| xAI | SpaceXAI: Grok 4.3 | $1.25 | $2.5 | $64.84 | source 2026-10-16 ✔ matches |
| xAI | SpaceXAI: Grok 4.20 Multi-Agent | $1.25 | $2.5 | $64.84 | source 2026-10-16 ✔ matches |
| xAI | SpaceXAI: Grok 4.20 | $1.25 | $2.5 | $64.84 | source 2026-10-16 ✔ matches |
| Google: Nano Banana 2.1 | $1.5 | $7.5 | $101.75 | source 2026-10-16 no 2nd source | |
| Mistral | Mistral: Mistral Medium 3.5 | $1.5 | $7.5 | $101.75 | source 2026-10-16 ✔ matches |
| Google: Gemini 3.5 Flash | $1.5 | $9.0 | $109.72 | source 2026-10-16 ✔ matches | |
| xAI | SpaceXAI: Grok 4.7 | $2.0 | $6.0 | $114.38 | source 2026-10-16 ✔ matches |
| xAI | SpaceXAI: Grok 4.6 | $2.0 | $6.0 | $114.38 | source 2026-10-16 ✔ matches |
| xAI | SpaceXAI: Grok 4.5 | $2.0 | $6.0 | $114.38 | source 2026-10-16 ✔ matches |
| Google: Gemini 4 Argon | $2.0 | $10.0 | $135.66 | source 2026-10-16 no 2nd source | |
| OpenAI | OpenAI: GPT-6.1 Sol Pro | $2.0 | $10.0 | $135.66 | source 2026-10-16 ✔ matches |
| OpenAI | OpenAI: GPT-6.1 Sol | $2.0 | $10.0 | $135.66 | source 2026-10-16 ✔ matches |
| Anthropic | Anthropic: Claude Sonnet 5.5 | $2.0 | $10.0 | $135.66 | source 2026-10-16 ✔ matches |
| OpenAI | OpenAI: GPT-6 Sol Pro | $2.0 | $10.0 | $135.66 | source 2026-10-16 ✔ matches |
| OpenAI | OpenAI: GPT-6 Sol | $2.0 | $10.0 | $135.66 | source 2026-10-16 ✔ matches |
| Anthropic | Anthropic: Claude Sonnet 5 | $2.0 | $10.0 | $135.66 | source 2026-10-16 ✔ matches |
| Anthropic | Anthropic: Claude Opus 5.5 | $4.0 | $20.0 | $271.32 | source 2026-10-16 ✔ matches |
| Anthropic | Anthropic: Claude Opus 5 | $5.0 | $25.0 | $339.15 | source 2026-10-16 ✔ matches |
| Anthropic | Anthropic: Claude Opus 4.8 | $5.0 | $25.0 | $339.15 | source 2026-10-16 ✔ matches |
| OpenAI | OpenAI: GPT-6 Astra | $10.0 | $50.0 | $678.30 | source 2026-10-16 ✔ matches |
| OpenAI | OpenAI: GPT-6 Astra Pro | $10.0 | $50.0 | $678.30 | source 2026-10-16 ✔ matches |
| Anthropic | Anthropic: Claude Fable 5.1 | $10.0 | $50.0 | $678.30 | source 2026-10-16 ✔ matches |
| Anthropic | Anthropic: Claude Fable 5 | $10.0 | $50.0 | $678.30 | source 2026-10-16 ✔ matches |
Try the interactive calculators with your own numbers →
Every week: which AI API prices changed, and what it means for your app's bill. Free.
Prices from OpenRouter's public API, cross-checked with InferIndex. No spam.