Cheapest AI API for summarising documents
We never take money for a place in a ranking. Read how we make money.
The verdict
- Scores 96 on overall, #2 of 31 models we track for this job.
- $11.00 per 1,000 summaries.
- Reads up to 1,049K tokens at once.
- Scores 55 on overall, #16 of 31 models we track for this job.
- $0.75 per 1,000 summaries, 93% less than the best overall pick.
- The cheapest model that passes our test for summarising documents cheaply.
- Scores 97 on overall, the highest of 31 models we track for this job.
- $22.00 per 1,000 summaries: the extra $11.00 per 1,000 buys 1 more points.
- Reads up to 1,000K tokens at once.
Summaries are the classic bulk job: lots of text in, a short answer out. Input price matters far more than output price here, which reshuffles the usual rankings.
How we tested
We use the overall score as the quality bar until our rig runs summary tasks. A summary is about 4,000 tokens in and 300 out, so the cost table is driven by input prices.
Scores are provisional until our own test rig runs these tasks. A model “passes our test” for this job at a score of 45 or more.
Scores and cost at your volume
| Model | Score | Passes | 1k summaries a month | 10k summaries a month | 100k summaries a month |
|---|---|---|---|---|---|
| Claude Opus 5.5 | 97 | Yes | $22.00 | $220 | $2200 |
| Gemini 4 Argon | 96 | Yes | $11.00 | $110 | $1100 |
| Claude Fable 5.1 | 92 | Yes | $55.00 | $550 | $5500 |
| Claude Opus 5 | 86 | Yes | $27.50 | $275 | $2750 |
| Muse Wonder | 80 | Yes | $22.00 | $220 | $2200 |
| Claude Sonnet 5.5 | 79 | Yes | $11.00 | $110 | $1100 |
| GPT-6 Astra | 74 | Yes | $55.00 | $550 | $5500 |
| GPT-6.1 Sol | 74 | Yes | $11.00 | $110 | $1100 |
| Gemini 3.8 Flash | 71 | Yes | $4.13 | $41.25 | $413 |
| Muse Spark 1.3 | 68 | Yes | $6.28 | $62.75 | $628 |
| GPT-5.6 Sol | 68 | Yes | $22.00 | $220 | $2200 |
| Qwen3.8 Max | 66 | Yes | $9.80 | $98.00 | $980 |
One summary is about 4,000 tokens in and 300 out, at list prices. Try your own numbers in the cost calculator.
Who should pick something else
- For legal or medical summaries where a missed detail is costly, use the research guide’s pick instead.
- If you summarise the same documents often, check prompt-caching prices in the cost calculator.
Questions
What is the cheapest model for summarisation?
The budget pick below: the cheapest model whose overall score clears our bar. It changes when prices move, and we update this page when it does.
Why does input price matter so much?
A summary reads ten times more than it writes, so the input price makes up most of the bill.
Changes to this guide
- 25 Oct 2026: First version, on provisional scores from Arena and Epoch AI.