Reports / LLM API Price Index / October 2026
Monthly report · issue 1 · data Oct 23, 2026
LLM API Price Index, October 2026
Short answerThe median listed LLM API model costs $7.00 for 1,000 chatbot replies (1,500 tokens in, 400 out) as of Oct 23, 2026, across 67 models from 4 providers. Index: 100.0 (October 2026 = 100). The cheapest is GPT-5 nano at $0.23; the most expensive, o1-pro, costs $465. 3 price changes are already announced for the coming months.
By provider
| Provider | Listed models | Cheapest | Median | Most expensive | Median agent step x1,000 |
|---|---|---|---|---|---|
| Anthropic | 14 | $0.35 | $15.75 | $35.00 | $225 |
| OpenAI | 41 | $0.23 | $7.75 | $465 | $100 |
| 10 | $0.31 | $2.29 | $7.80 | $31.75 | |
| DeepSeek | 2 | $0.93 | $2.25 | $3.56 | $37.56 |
Download this issue as CSV · every model: /llm/prices.csv
Changes since the last issue
This is the first issue, so there is nothing to compare yet. Prices were first recorded on Oct 17, 2026, Oct 23, 2026.
- No listed price changed between our checks this month.
Announced for later
- Dec 1, 2026: Gemini 4 Argon to $4 input / $20 output (now $2 / $10).
- Jan 1, 2027: Gemini 3.8 Flash to $1.5 input / $7.5 output (now $0.75 / $3.75).
- Jan 1, 2027: Gemini 3.6 Flash to $1.5 input / $7.5 output (now $0.75 / $3.75).
Method
Basket: every model the provider lists as current on its official pricing page with an input and an output price (retired models and embeddings excluded), from our dated snapshot of Oct 23, 2026. Workload: one chatbot reply, 1,500 input tokens and 400 output tokens, at the standard price (no cache or batch discount); the agent column is a 40,000-in, 2,000-out coding-agent step. The index is the median basket cost divided by the first issue's median. A model joining or leaving the basket moves the index, which is the point: it tracks what a buyer can actually pick. Per-model detail: LLM API prices; every recorded change: changelog.