AI task costs · in pounds
What it costs to search and answer questions over 10,000 internal documents with AI
Answering questions: gpt-oss-120b is the cheapest of the 24 models priced, at 32p for 2,000 questions a month (0.016p per question). The dearest, GPT-6 Astra, costs £88.19; the middle of the range is Mistral Large 4 at £5.40.
How much does it cost to let staff ask questions of company documents with AI (retrieval and embeddings)? The answer below covers price only, not quality.
Prices of from OpenRouter's public API price list · converted at the ECB reference rate of ($1 = £0.7538) · Urbano DX
Answering questions
The workload
- A search finds the passages most relevant to a question; the model reads them and answers.
- Per question: 50 tokens of question, 300 tokens of instructions and 4,000 tokens of passages: 4,350 input tokens.
- An answer of about 220 words with its sources: 300 output tokens.
- 2,000 questions a month.
| Model | Per question | 2,000 questions a month | Rate used (USD) |
|---|---|---|---|
| gpt-oss-120bOpenAI | 0.016p | 32p | $0.037 in, $0.17 out per 1M tokens |
| DeepSeek V4.1 FlashDeepSeek | 0.027p | 54p | $0.02016 in, $0.891 out per 1M tokens |
| Gemma 4 31BGoogle | 0.037p | 74p | $0.09 in, $0.34 out per 1M tokens |
| Claude Haiku 5.5Anthropic | 0.044p | 88p | $0.10 in, $0.50 out per 1M tokens |
| GPT-6 LunaOpenAI | 0.044p | 88p | $0.10 in, $0.50 out per 1M tokens |
| Qwen3.8 FlashQwen | 0.06p | £1.20 | $0.15 in, $0.47 out per 1M tokens |
| Mistral Small 4Mistral | 0.063p | £1.26 | $0.15 in, $0.60 out per 1M tokens |
| Gemini 2.5 FlashGoogle | 0.15p | £3.10 | $0.30 in, $2.50 out per 1M tokens |
| Gemini 3.5 Flash LiteGoogle | 0.15p | £3.10 | $0.30 in, $2.50 out per 1M tokens |
| GPT-4.1 MiniOpenAI | 0.17p | £3.35 | $0.40 in, $1.60 out per 1M tokens |
| Llama 3.3 70B InstructMeta | 0.21p | £4.23 | $0.59 in, $0.79 out per 1M tokens |
| Mistral Large 4Mistral | 0.27p | £5.40 | $0.68 in, $2.09 out per 1M tokens |
| Gemini 3.8 FlashGoogle | 0.33p | £6.61 | $0.75 in, $3.75 out per 1M tokens |
| DeepSeek V4 Pro 0423DeepSeek | 0.36p | £7.13 | $0.9553 in, $1.911 out per 1M tokens |
| Claude Haiku 4.5Anthropic | 0.44p | £8.82 | $1.00 in, $5.00 out per 1M tokens |
| Muse Spark 1.3Meta | 0.51p | £10.12 | $1.25 in, $4.25 out per 1M tokens |
| Grok 4.7SpaceXAI | 0.79p | £15.83 | $2.00 in, $6.00 out per 1M tokens |
| Qwen3.8 Max (0902)Qwen | 0.79p | £15.83 | $2.00 in, $6.00 out per 1M tokens |
| Claude Sonnet 5.5Anthropic | 0.88p | £17.64 | $2.00 in, $10.00 out per 1M tokens |
| Gemini 4 ArgonGoogle | 0.88p | £17.64 | $2.00 in, $10.00 out per 1M tokens |
| GPT-6.1 SolOpenAI | 0.88p | £17.64 | $2.00 in, $10.00 out per 1M tokens |
| Claude Opus 5.5Anthropic | 1.8p | £35.28 | $4.00 in, $20.00 out per 1M tokens |
| Claude Fable 5.1Anthropic | 4.4p | £88.19 | $10.00 in, $50.00 out per 1M tokens |
| GPT-6 AstraOpenAI | 4.4p | £88.19 | $10.00 in, $50.00 out per 1M tokens |
Indexing the documents (embeddings)
The workload
- 10,000 documents of about 1,100 words, 1,500 tokens each: 15 million tokens to embed once.
- Each month 5% of the documents change (750,000 tokens) and each of the 2,000 questions is embedded too (100,000 tokens): 850,000 tokens a month.
- Storing the vectors and running the search are not included.
| Model | Index all 10,000 once | Updates and questions, a month | Rate used (USD) |
|---|---|---|---|
| bge-m3BAAI | 11p | 0.64p | $0.01 per 1M tokens |
| Multilingual-E5-LargeIntfloat | 11p | 0.64p | $0.01 per 1M tokens |
| Qwen3 Embedding 8BQwen | 11p | 0.64p | $0.01 per 1M tokens |
| Qwen3 Embedding 4BQwen | 23p | 1.3p | $0.02 per 1M tokens |
| Text Embedding 3 SmallOpenAI | 23p | 1.3p | $0.02 per 1M tokens |
| voyage-4-liteVoyageAI by MongoDB | 23p | 1.3p | $0.02 per 1M tokens |
| Embed V1 4BPerplexity | 34p | 1.9p | $0.03 per 1M tokens |
| voyage-4VoyageAI by MongoDB | 68p | 3.8p | $0.06 per 1M tokens |
| Mistral Embed 2312Mistral | £1.13 | 6.4p | $0.10 per 1M tokens |
| voyage-4-largeVoyageAI by MongoDB | £1.36 | 7.7p | $0.12 per 1M tokens |
| Text Embedding 3 LargeOpenAI | £1.47 | 8.3p | $0.13 per 1M tokens |
| Gemini Embedding 001Google | £1.70 | 9.6p | $0.15 per 1M tokens |
| Gemini Embedding 2Google | £2.26 | 13p | $0.20 per 1M tokens |
How the numbers are worked out
- Each workload is written out on its page: tokens in and out, the share read from a prompt cache, audio minutes or images. Cost = workload × the model's listed rate, so you can check any figure with a calculator.
- Rates are OpenRouter's API prices in our snapshot of , the same ones on our model pricing pages. Buying from the developer directly can cost a little more or less. When rates move, the price change log records it.
- US dollars become pounds at the European Central Bank reference rate of : €1 = $1.1195 = £0.84385, so $1 = £0.7538. VAT, card fees and top-up fees are not included.
- About 0.75 English words make one token. Each model's tokenizer counts the same text a little differently, so treat a figure as a close estimate, not a quote.
- Models that reason before answering can add thinking tokens, billed as output. These figures count none, so a model with thinking switched on can cost more.
- This compares price only. It says nothing about which model does the job well: a cheap model that gets a tenth of the answers wrong can cost more in staff time than it saves.
Other tasks
Want this running on your own data, with the cost measured? See the fixed-price AI pilot