As of October 14, 2026, the cheapest major LLM API on standard rates is GPT-6 Luna (OpenAI) and Claude Haiku 5.5 (Anthropic), tied at $0.10 in / $0.50 out per 1M tokens. Among the models people compare most, Grok 4.7 ($2 in / $6 out per 1M tokens) is cheaper than Claude Opus 5.5 ($4 in / $20 out per 1M tokens). Enter your own token volumes in the calculator below for the real answer for your workload.
LLM API prices per 1M tokens (October 14, 2026)
Standard (non-cached, non-batch) published rates, sorted cheapest first by a blended 3:1 input:output mix. Last verified October 14, 2026.
| Model | Provider | Input / 1M | Output / 1M | Tier |
|---|---|---|---|---|
| GPT-6 Luna | OpenAI | $0.10 | $0.50 | Budget |
| Claude Haiku 5.5 | Anthropic | $0.10 | $0.50 | Budget |
| DeepSeek V4.1 Flash | DeepSeek | $0.15 | $0.60 | Budget |
| Mistral Small 4 | Mistral | $0.15 | $0.60 | Budget |
| GPT-5.6 Luna | OpenAI | $0.20 | $1.20 | Budget |
| Mistral Large 3 | Mistral | $0.50 | $1.50 | Budget |
| DeepSeek V4 Pro | DeepSeek | $0.66 | $1.98 | Budget |
| Mistral Large 4 | Mistral | $0.68 | $2.09 | Budget |
| Gemini 3.8 Flash | $0.75 | $3.75 | Balanced | |
| Claude Haiku 4.5 | Anthropic | $1 | $5 | Budget |
| Grok 4.7 | xAI | $2 | $6 | Balanced |
| GPT-5 | OpenAI | $1.25 | $10 | Frontier |
| GPT-6.1 Sol | OpenAI | $2 | $10 | Balanced |
| Claude Sonnet 5.5 | Anthropic | $2 | $10 | Balanced |
| GPT-5.6 Terra | OpenAI | $2 | $12 | Balanced |
| Gemini 3.1 Pro | $2 | $12 | Frontier | |
| Claude Opus 5.5 | Anthropic | $4 | $20 | Frontier |
| Meta Muse Wonder | Meta | $4 | $20 | Frontier |
| Claude Opus 5 | Anthropic | $5 | $25 | Frontier |
| GPT-6 Astra | OpenAI | $10 | $50 | Frontier |
| Claude Fable 5.1 | Anthropic | $10 | $50 | Frontier |
Prices change often; batch APIs and cached-input reads can cut these further. Verify with each provider before committing.
How much will your LLM API bill be?
Estimate and compare the monthly cost of running GPT, Claude, Gemini, Grok, DeepSeek, and Mistral for your workload — then see how much prompt compression could shave off. Prices are per 1M tokens, as of October 14, 2026.
New here? Read the full guide: how to estimate your LLM API costs| Model | $/1M in | $/1M out | Cost / request | Monthly | |
|---|---|---|---|---|---|
GPT-6 LunaBudget OpenAI | $0.10 | $0.50 | 0.120¢ | $60.00 | Details |
Claude Haiku 5.5Budget Anthropic | $0.10 | $0.50 | 0.120¢ | $60.00 | Details |
DeepSeek V4.1 FlashBudget DeepSeek | $0.15 | $0.60 | 0.168¢ | $84.00 | Details |
Mistral Small 4Budget Mistral | $0.15 | $0.60 | 0.168¢ | $84.00 | Details |
GPT-5.6 LunaBudget OpenAI | $0.20 | $1.20 | 0.256¢ | $128.00 | Details |
Mistral Large 3Budget Mistral | $0.50 | $1.50 | 0.520¢ | $260.00 | Details |
DeepSeek V4 ProBudget DeepSeek | $0.66 | $1.98 | 0.686¢ | $343.20 | Details |
Mistral Large 4Budget Mistral | $0.68 | $2.09 | 0.711¢ | $355.60 | Details |
Gemini 3.8 FlashBalanced Google | $0.75 | $3.75 | 0.900¢ | $450.00 | Details |
Claude Haiku 4.5Budget Anthropic | $1.00 | $5.00 | $0.01 | $600.00 | Details |
GPT-5Frontier OpenAI | $1.25 | $10.00 | $0.02 | $900.00 | Details |
Grok 4.7Balanced xAI | $2.00 | $6.00 | $0.02 | $1,040 | Details |
GPT-6.1 SolBalanced OpenAI | $2.00 | $10.00 | $0.02 | $1,200 | Details |
Claude Sonnet 5.5Balanced Anthropic | $2.00 | $10.00 | $0.02 | $1,200 | Details |
GPT-5.6 TerraBalanced OpenAI | $2.00 | $12.00 | $0.03 | $1,280 | Details |
Gemini 3.1 ProFrontier Google | $2.00 | $12.00 | $0.03 | $1,280 | Details |
Claude Opus 5.5Frontier Anthropic | $4.00 | $20.00 | $0.05 | $2,400 | Details |
Meta Muse WonderFrontier Meta | $4.00 | $20.00 | $0.05 | $2,400 | Details |
Claude Opus 5Frontier Anthropic | $5.00 | $25.00 | $0.06 | $3,000 | Details |
GPT-6 AstraFrontier OpenAI | $10.00 | $50.00 | $0.12 | $6,000 | Details |
Claude Fable 5.1Frontier Anthropic | $10.00 | $50.00 | $0.12 | $6,000 | Details |
Estimates only. Prices are standard (non-cached, non-batch) per-1M-token rates as of October 14, 2026 and change frequently — verify with each provider. Batch and cached-input discounts can cut these further.
Cut the bill with prompt compression
Most of an LLM bill is input tokens — long chat histories, retrieved docs, and tool traces. Prompt compression removes low-value context before the call. See what a reduction does to your input spend.
Picking between these models?
Compare them side by side on features, pricing, and real pros & cons.
How the LLM cost calculator works
Every API call bills separately for the tokens you send (input) and the tokens the model generates (output), each priced per million tokens. This calculator multiplies your input and output tokens per request by each model's rate, then by your monthly request volume, so you can compare providers on the exact workload you run. A token is roughly 4 characters, or about ¾ of a word.
Full walkthrough: how to estimate and cut your LLM API costsWhy is my output so much more expensive than input?
Output tokens usually cost 4–8× more than input tokens across providers. If your app generates long answers, output dominates the bill; if it sends huge context (RAG, long chats), input dominates — which is where prompt compression helps most.
Are these prices current?
They reflect standard published rates as of October 14, 2026. LLM pricing moves fast — providers cut prices and launch new tiers regularly — so always confirm on the provider's pricing page before budgeting. Batch APIs and cached-input reads can lower costs well below the figures shown here.
Frequently asked questions
What is the cheapest LLM API right now?
As of October 14, 2026, the cheapest major LLM API on standard rates is GPT-6 Luna (OpenAI) and Claude Haiku 5.5 (Anthropic), tied at $0.10 in / $0.50 out per 1M tokens. Among independent and open-weight options, DeepSeek V4.1 Flash ($0.15 in / $0.60 out) and Mistral Small 4 ($0.15 in / $0.60 out) are the lowest. The cheapest choice for you depends on your input:output mix — use the calculator below with your own token numbers.
How much does the Claude API cost per million tokens?
As of October 14, 2026, Anthropic's Claude API standard rates are: Claude Opus 5.5 at $4 in / $20 out per 1M tokens; Claude Sonnet 5.5 at $2 in / $10 out per 1M tokens; Claude Haiku 4.5 at $1 in / $5 out per 1M tokens; and Claude Haiku 5.5 at $0.10 in / $0.50 out per 1M tokens. Batch and cached-input reads cost less. Verify current rates on Anthropic's pricing page.
Which is cheaper, Grok 4.7 or Claude Opus 5.5?
Grok 4.7 is cheaper. As of October 14, 2026, Grok 4.7 (xAI) is $2 in / $6 out per 1M tokens, versus Claude Opus 5.5 at $4 in / $20 out per 1M tokens — roughly half the input price and under a third of the output price. They target different jobs, though: Opus 5.5 is a frontier model, while Grok 4.7 sits in the balanced tier, so compare on quality for your task as well as price.