Telegraph-style compression test shows accuracy and cost tradeoffs
What happened
On October 7, Hacker News featured the Telegraph Test, an evaluation of telegraph-style compression for LLMs. It covered 50 texts and about 1,300 questions. Across models, reading compressed text gave accuracy ratios of 0.99 to 1.10 versus a plaintext baseline. With lowercase instructions, GLM-5.3-Flash records saved 48.4% of tokens, but answering directly in telegraph style dropped the accuracy ratio to 0.81. On cost, gpt-5-mini cannot turn off reasoning, so compressed writes cost about twice as much as plaintext, showing fewer record tokens does not always mean cheaper writes.
Written by AI from the coverage · updated 40 minutes ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pagePickTelegraph Test: LLMs compress text into telegraph style, and reading it back stays accurate
Telegraph Test measured LLM telegraph-style compression across 50 texts and about 1,300 questions. Accuracy when another model reads the compressed text ran 0.99 to 1.10 times the plaintext baseline. With lowercase instructions, GLM-5.3-Flash cut tokens by 48.4% but answered in telegraph style at an accuracy ratio of 0.81. gpt-5-mini, which cannot turn off reasoning, cost about twice as much to write the compressed record as plaintext.
Heat over time
Not enough continuous observations to draw a trend yet.