Which LLM is best for polishing academic writing?

Judge by measured pass rate on academic tasks (polish + LaTeX correction) in a live GPT Academic instance, then take the cheapest full-pass model. Conservative rewriting that preserves meaning matters more than eloquence.

学术润色该用哪个大模型？

看真实 GPT Academic 实例上润色 + LaTeX 纠错两项任务的实测通过率，在全过的模型里选最便宜的。保留原意的保守改写比文采更重要。

Measured evidence: gpt-academic

Updated 2026-08-01; 4 models measured, 3 passed every task; the lowest measured cost among complete passes was gpt-5.6-luna at $0.008296 per benchmark run.

Human-readable benchmark · Raw JSON data

Agent workflow

  1. Read the gpt-academic live benchmark.
  2. Choose a full-pass model, cheapest first.
  3. Run your own paragraph through ask_model.
  4. Compare two models on the same paragraph before batch work.

Recommended tool: ask_model

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "ask_model",
    "arguments": {
      "prompt": "Polish for an academic journal, keep meaning: \"The results shows that our method perform better than baseline in most case.\""
    }
  }
}

MCP endpoint: https://ainetcafe.com/mcp · Machine-readable guide

Method and reuse

Live application tasks are run repeatedly; pass/fail, actual billed USD cost, and latency are recorded. Upstream-unavailable models are excluded from quality conclusions.

Data license: CC BY 4.0. Cite the page URL and measurement date.

Updated 2026-08-01 · All benchmark data · Agent capability index