Document summarisation: what it costs and which AI to use
Turn reports, contracts, articles or emails into a short summary or a set of key points. Operations, legal, research and finance teams processing a steady stream of long documents.
The short answer
At 2,000 documents a month, we estimate $1.86 on DeepSeek V4 Flash 0423 (the cheapest model that cleared our quality bar), $10.80 on Mistral Large 3 2512, and $126 on GPT-5.5. The flagship costs 12x the middle pick. Judged on summarising and structured-extraction tasks: missed facts and made-up facts are the failures.
Simulate your document summarisation
Pick a job, set how much of it you do, and see what every model would cost. Defaults are typical, not yours. Change them.
Prompt, history, documents
What the model writes
Advanced: caching, reasoning, retries, discounts
Share that can use cached-input prices where a model offers them
1 = none. 2 = hidden thinking doubles billed output
Extra calls on top of the base count
Batch jobs or committed spend. Check your provider; we assume none by default
Lowest benchmark score a model can have and still be picked. Higher is stricter.
Limit to certain providers
Untick any provider you cannot use, for data-residency, contract or policy reasons. Picks are recalculated.
2,000 documents a month, three ways to pay
Every model here cleared a score of 9.0 out of 10 on our summarise, extract json benchmark tasks, ran without errors and fits a 32K-token window. 21 of 386 priced models qualify.
Cheapest model that clears the bar.
DeepSeek V4 Flash 0423
- Per document
- $0.0009
- Benchmark
- 9.50 / 10 · 6 prompts
- Context
- 1.05M tokens
Priced near the middle of the models that clear the bar; best score there.
Mistral Large 3 2512
- Per document
- $0.0054
- Benchmark
- 9.50 / 10 · 6 prompts
- Context
- 262K tokens
What many people reach for first.
GPT-5.5
- Per document
- $0.063
- Benchmark
- 9.50 / 10 · 6 prompts
- Context
- 1.05M tokens
Choosing GPT-5.5 over Mistral Large 3 2512 costs $115 more a month, $1,382 a year (12x the price).
Both scored 9.50 out of 10 on this kind of work. The extra money buys safety margin our benchmark cannot measure: it is 24 short prompts, and most passing models score near the top. If a wrong answer is costly for you, that margin can be worth it. Test on your own data before you decide.
If it grows
Pay-per-use cost scales in a straight line. Monthly cost at 1x, 3x, 10x and 100x today's volume.
| Model | 1x · 2,000 | 3x · 6,000 | 10x · 20K | 100x · 200K |
|---|---|---|---|---|
| DeepSeek V4 Flash 0423 | $1.86 | $5.57 | $18.56 | $186 |
| Mistral Large 3 2512 | $10.80 | $32.40 | $108 | $1,080 |
| GPT-5.5 | $126 | $378 | $1,260 | $12,600 |
What a budget buys
- DeepSeek V4 Flash 0423108Kdocuments a month, about 3,591 a day
- Mistral Large 3 251219Kdocuments a month, about 617 a day
- GPT-5.51,587documents a month, about 52 a day
Every model, priced for your workload
41 models, 15 hidden because their window is smaller than 32K tokens.
| Model | A month | Per document | Benchmark | Context |
|---|---|---|---|---|
| DeepSeek V4.1 Flash DeepSeekbelow the bar | $1.41 | $0.0007 | 8.33 (6) | 1.05M |
| DeepSeek V4 Flash 0423 DeepSeekBudget pick | $1.86 | $0.0009 | 9.50 (6) | 1.05M |
| Llama 4 Scout Meta | $2.16 | $0.0011 | 9.50 (6) | 1.31M |
| Ministral 3 8B 2512 Mistral AI | $2.88 | $0.0014 | 9.33 (6) | 262K |
| GLM 5.3 Z Aibelow the bar | $2.94 | $0.0015 | 8.67 (6) | 1.05M |
| Qwen3.8 Flash Alibababelow the bar | $3.26 | $0.0016 | 8.17 (6) | 1M |
| GLM 5.3 Flash Z Ai | $3.30 | $0.0016 | 9.33 (6) | 1.05M |
| Mistral Small 4 Mistral AI | $3.42 | $0.0017 | 9.50 (6) | 262K |
| Qwen3.8 27B Alibababelow the bar | $4.95 | $0.0025 | 5.83 (6) | 1.02M |
| GPT-5.4 Nano OpenAI | $5.10 | $0.0026 | 9.50 (6) | 400K |
| MiniMax M2.5 Minimax | $6.00 | $0.0030 | 9.33 (6) | 205K |
| Muse Glimmer 30B Metabelow the bar | $6.84 | $0.0034 | 8.50 (6) | 131K |
Benchmark = mean judge score out of 10 on the summarise, extract json prompts of the Model Playground, with the number of prompts in brackets. Models marked n/a have no benchmark run, so we do not guess their quality. How the benchmark works.
What we assumed
- An average document is about 6,500 words, roughly 9,000 tokens.
- The summary is about 450 words, roughly 600 tokens.
- No caching, because every document is different.
These are typical numbers, not measurements of your business. Edit them in the simulator above.
What drives the cost
Input dominates: the document is long and the summary is short, so the input price per million tokens is what you pay.
Watch out for
- Check the context window. A 100-page contract can run past 100,000 tokens, which rules out some cheap models.
- If you summarise the same document several ways, cache it. The second read is much cheaper on models that support caching.
Questions people ask
How much will document summarisation cost per month?
At 2,000 documents a month, we estimate $1.86 a month on DeepSeek V4 Flash 0423, the cheapest model that clears our quality bar for this job. Mistral Large 3 2512 costs $10.80 and GPT-5.5 costs $126. These are estimates from the assumptions listed on this page, not a quote. Change the numbers to match your own volume.
What is the cheapest AI model for document summarisation?
On current OpenRouter prices, DeepSeek V4 Flash 0423 from DeepSeek is the cheapest model that cleared our bar of 9.0 out of 10, scoring 9.5/10 on 6 benchmark prompts for this kind of work. The benchmark is small, so treat it as a shortlist and test on your own data before switching.
Is it worth paying for a flagship model for document summarisation?
On our benchmark, GPT-5.5 scored 9.5/10 on this kind of work, and a model priced near the middle of the models that cleared the bar, Mistral Large 3 2512, scored 9.5/10, while the flagship costs 12x as much. A flagship buys safety margin the benchmark cannot measure. If a wrong answer is expensive for you, that can be worth paying for. If not, you are probably overpaying.
What happens to the cost if document summarisation grows ten times?
Pay-per-use pricing scales in a straight line, so ten times the documents is ten times the bill. Use the growth table on this page to see 3x, 10x and 100x for the same three models.
What makes document summarisation expensive?
Input dominates: the document is long and the summary is short, so the input price per million tokens is what you pay.
Price a different job
- Customer support chatbotFrom $1.74 a month at 3,000 conversations
- Coding assistant for a teamFrom $4.12 a month at 4,400 requests
- Blog and content writingFrom $0.083 a month at 120 articles
- Knowledge base Q&A (RAG)From $5.62 a month at 15K questions
- Invoice and form data extractionFrom $4.65 a month at 20K documents
- Ticket and content classificationFrom $11.30 a month at 500K items
- AI agent workflowsFrom $7.44 a month at 1,500 runs
- Sales email personalisationFrom $0.595 a month at 5,000 emails
- Meeting notes and summariesFrom $0.575 a month at 400 meetings
- Code review for pull requestsFrom $0.840 a month at 600 pull requests
- Personal AI assistant for a founderFrom $0.478 a month at 1,500 messages
- Text-to-SQL data analystFrom $2.54 a month at 6,000 queries