Cheapest LLM for high-volume customer support
At a million replies a month, a fraction of a cent per reply becomes a real line item. This ranks every model on a support workload with a cached instruction and policy prompt. Prices verified 2026-10-18.
Assumptions
- 2,500 input tokens per reply (instructions, policies, conversation), 60% of it served from cache
- 300 output tokens per reply
- 1,000,000 replies per month, standard (real-time) pricing
Change any of these in the calculator: every row below has a link that opens the same scenario for that model.
The short answer
- Cheapest flagship model: Mistral Large 4 at $1.41 (temporary sale price; regular $1.36 / $4.18 per 1M)
- Cheapest mid-tier model: Gemini 3.8 Flash at $1.99 (price rises 2027-01-01)
- Cheapest small model: Ministral 3 8B at $0.22
Small model plus escalation
Many teams answer routine questions with a small model and escalate hard tickets. Sending 90% of replies to Ministral 3 8B and 10% to Mistral Large 4 costs $337 per month, compared with $218 for the small model alone and $1,412 for the flagship alone. The routing step itself (a classifier call or rules) adds a little on top.
Caching the policy prompt
For Ministral 3 8B, serving 60% of the prompt from cache brings the month from $420 to $218. Keep the stable part (instructions, policies, examples) at the start of the prompt and the customer's message at the end so the prefix stays identical.
Trade-offs
Support quality is measured in resolved tickets and avoided escalations, not tokens. A model that resolves 5% more tickets without a human usually saves far more than the token bill difference. Track resolution rate and customer satisfaction per model before cutting cost.
Full ranking (36 models)
| # | Model | Tier | Per 1,000 replies | Per month (1M replies) | |
|---|---|---|---|---|---|
| 1 | Ministral 3 8B Mistral AI | small | $0.22 | $218 | Calculator |
| 2 | Claude Haiku 5.5 Anthropic | small | $0.27 | $265 | Calculator |
| 3 | GPT-6 Luna OpenAI | small | $0.27 | $265 | Calculator |
| 4 | Mistral Small 4 Mistral AI | small | $0.35 | $353 | Calculator |
| 5 | GPT-5.6 Luna OpenAI | small | $0.59 | $590 | Calculator |
| 6 | Codestral Mistral AI | small | $0.61 | $615 | Calculator |
| 7 | DeepSeek V4.1 Flash DeepSeek | small | $0.67 | $669 | Calculator |
| 8 | Gemini 3.1 Flash-Lite Google | small | $0.74 | $738 | Calculator |
| 9 | Gemini 3.5 Flash-Lite Google | small | $1.09 | $1,095 | Calculator |
| 10 | Gemini 2.5 Flash Google | small | $1.09 | $1,095 | Calculator |
| 11 | Mistral Large 4 Mistral AI sale price | flagship | $1.41 | $1,412 | Calculator |
| 12 | Grok Build 0.1 xAI | small | $1.90 | $1,900 | Calculator |
| 13 | Gemini 3.8 Flash Google price changes 2027-01-01 | mid | $1.99 | $1,988 | Calculator |
| 14 | Gemini 3.6 Flash Google price changes 2027-01-01 | mid | $1.99 | $1,988 | Calculator |
| 15 | GPT-5.4 mini OpenAI | small | $2.21 | $2,213 | Calculator |
| 16 | Grok 4.20 (Reasoning) xAI | reasoning | $2.30 | $2,300 | Calculator |
| 17 | Grok 4.3 xAI | mid | $2.30 | $2,300 | Calculator |
| 18 | DeepSeek V4 Pro DeepSeek | flagship | $2.57 | $2,574 | Calculator |
| 19 | Claude Haiku 4.5 Anthropic | small | $2.65 | $2,650 | Calculator |
| 20 | Mistral Medium 3.5 Mistral AI | mid | $3.98 | $3,975 | Calculator |
| 21 | Grok 4.7 xAI | flagship | $4.55 | $4,550 | Calculator |
| 22 | Gemini 4 Argon Google price changes 2026-12-01 | flagship | $5.15 | $5,150 | Calculator |
| 23 | Claude Sonnet 5.5 Anthropic | mid | $5.15 | $5,150 | Calculator |
| 24 | GPT-6.1 Sol OpenAI | mid | $5.15 | $5,150 | Calculator |
| 25 | Claude Sonnet 5 Anthropic | mid | $5.30 | $5,300 | Calculator |
| 26 | GPT-6 Sol OpenAI | mid | $5.30 | $5,300 | Calculator |
| 27 | Gemini 3.1 Pro (Preview) Google | flagship | $5.90 | $5,900 | Calculator |
| 28 | GPT-5.6 Terra OpenAI | mid | $5.90 | $5,900 | Calculator |
| 29 | Claude Sonnet 4.6 Anthropic | mid | $7.95 | $7,950 | Calculator |
| 30 | Claude Opus 5.5 Anthropic | flagship | $10.30 | $10,300 | Calculator |
| 31 | GPT-5.6 Sol OpenAI | flagship | $10.60 | $10,600 | Calculator |
| 32 | Claude Opus 5 Anthropic | flagship | $13.25 | $13,250 | Calculator |
| 33 | Claude Opus 4.8 Anthropic | flagship | $13.25 | $13,250 | Calculator |
| 34 | GPT-5.5 OpenAI | flagship | $14.75 | $14,750 | Calculator |
| 35 | Claude Fable 5.1 Anthropic | flagship | $25.38 | $25,375 | Calculator |
| 36 | GPT-6 Astra OpenAI | flagship | $26.50 | $26,500 | Calculator |
USD, standard list prices. Tags mark announced price changes and temporary sale prices.
Price is not quality
This page ranks by cost only. Models differ in accuracy, instruction following, tool use, latency, rate limits and data policies, and those differences often matter more than the price gap. Shortlist two or three models from the table, run them on a sample of your real inputs, and compare cost per successful result.
Related comparisons
More guides
- Cheapest LLM for coding agents
- Cheapest LLM for RAG chatbots
- Cheapest LLM for batch classification and extraction
- Long-context document analysis (400K tokens)
- Prompt caching explained, with savings
FAQ
How much does an AI support bot cost per month?
For 1,000,000 replies with the assumptions above, from $218 (Ministral 3 8B) to $26,500 (GPT-6 Astra).
What is the cost per support reply?
$0.00022 on Ministral 3 8B and $0.0014 on the cheapest flagship model, Mistral Large 4.