Verified November 2026

Home › Best › LLM for RAG

Best LLM for RAG

Ranked by the cost of one RAG request: 10,000 retrieved tokens (assumed 80% cached) plus 500 output tokens. Cached input is used where the provider publishes a cache rate, otherwise the standard input rate. Re-run with your own retrieval size in the calculator.

Top 15 for RAG (per request cost)

#ModelProviderCost / requestCached input $/MOutput $/MContext
1GPT-6 LunaOpenAI$0.0003$0.01$0.501,050,000
2Claude Haiku 5.5Anthropic$0.0003$0.01$0.50200,000
3Mistral Small 4Mistral AI$0.0004$0.015$0.60256,000
4DeepSeek V4.1 FlashDeepSeek$0.0007$0.006$1.201,100,000
5GPT-5.6 LunaOpenAI$0.0008$0.02$1.201,050,000
6GPT-5.4 nanoOpenAI$0.0008$0.02$1.25400,000
7Gemini 3.1 Flash-LiteGoogle$0.0010$0.025$1.501,000,000
8Gemini 3.5 Flash-LiteGoogle$0.0015$0.03$2.501,000,000
9Mistral Large 4Mistral AI$0.0018$0.07$2.101,000,000
10DeepSeek V4 ProDeepSeek$0.0024$0.044$3.961,000,000
11Gemini 3.8 FlashGoogle$0.0026$0.075$3.751,000,000
12Gemini 3.6 FlashGoogle$0.0026$0.075$3.751,000,000
13GPT-6.1 TerraOpenAI$0.0030$0.05$51,050,000
14GPT-5.4 miniOpenAI$0.0030$0.075$4.50400,000
15Grok 4.3xAI$0.0033$0.20$2.501,000,000

Frequently asked

What makes an LLM good for RAG?

A large context window for the retrieved chunks, a low cached-input rate because RAG prefixes are repeated, and a reasonable output price. Cache hits are where RAG workloads save the most.

How much does a RAG request cost?

At 10K retrieved tokens with an 80% cache hit, the cheapest models land around $0.001 per request and frontier reasoning models around $0.02. Multiply by your daily query volume to size the bill.

Should I use a long-context model for RAG?

Only if your retrieved set exceeds the standard window. Otherwise a 1M-context model is wasted money; a 128K model with cheap cached input is usually the better fit.

Browse all 45 models · Compare two models