TOON
Data format in Developer tools
People recommend it for cutting tokens when sending structured data to LLMs, since field names are declared once. Critics say savings vanish against optimized JSON.
People recommend it for cutting tokens when sending structured data to LLMs, since field names are declared once. Critics say savings vanish against optimized JSON.
23 people gave an opinion
- recommend it
- 13
- criticize it
- 10
- approval share
- 57%
Recommended in 8 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: not recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: not recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 13 people recommend itNeeds 10 or more, met.
- 1.3 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 8 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Fewer tokens than JSON
“TOON was much more compact, but when researchers examined whole-session effects, it was a wash”
Comment from 11 Aug 2026 in “What's the best programming language for coding agents?” “TOON's structural overhead is lower than compact JSON for the structure families analyzed here”
Comment from 28 Nov 2025 in “Toon vs. JSON: Byte-Level Efficiency Model” “TOON format: 50-58% token reduction for LLM interactions”
Comment from 5 Jan 2026 in “Show HN: RustAPI – An AI-first Rust API framework” Works well in practice
“TOON is a real game changer when used as an intermediary between the IDE/chat and the LLM”
Comment from 2 Dec 2025 in “My Contribution to Toon” “LLMs parse it correctly in our testing (GPT-4o, Claude, etc).”
Comment from 4 Feb 2026 in “We added TOON compression to our LLM gateway – compress prompts, saves tokens” “that works remarkably well too”
Comment from 5 Mar 2026 in “Show HN: XML, Markdown, or JSON: Which gives LLMs the most reliable boundaries?”
What people warn about
Each mark is one comment
Savings vanish in whole-session tests
“TOON was much more compact, but when researchers examined whole-session effects, it was a wash”
Comment from 11 Aug 2026 in “What's the best programming language for coding agents?” “TOON’s advertised 40–60% savings dropped to ~2% overall once JSON was fully optimized.”
Comment from 30 Nov 2025 on Hacker News “suggesting that TOON doesn't help overall”
Comment from 23 Jun 2026 in “The Token Compression Illusion: Why I'm Skeptical of RTK” Models unfamiliar with the format
“Toon is 3 months old, it's not used by anyone, and it's therefore not in the training set of any model.”
Comment from 17 Jan 2026 in “LLM Structured Outputs Handbook” “LLMs were less familiar with it, and thus spent even more tokens trying to decipher it”
Comment from 13 Feb 2026 in “Improving 15 LLMs at Coding in One Afternoon. Only the Harness Changed” “Is there evidence that LLMs adhere to this format better than to JSON? I doubt that.”
Comment from 17 Jan 2026 in “LLM Structured Outputs Handbook” Worse than JSON in some tests
“TOON performed worse than both JSON and having the LLM just consume the human output.”
Comment from 8 May 2026 in “Principles for agent-native CLIs” “It’s the same story we’ve seen before with TOON.”
Comment from 12 Jan 2026 in “Show HN: An LLM-optimized programming language”
What people recommend instead
The same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| JSON5 | 15 | 1 | 94% | 10 | — | Approved |
| TOONThis page | 13 | 10 | 57% | 8 | — | — |
What people said
22 of the 25 comments that mention it, newest first. Each links to the original comment.
“TOON never describes such behavior anywhere in its spec.”
“This TOON thing is just vibes.”
“TOON was much more compact, but when researchers examined whole-session effects, it was a wash”
“suggesting that TOON doesn't help overall”
“takes on a much more achievable goal - reduce whitespace and verbosity of JSON”
“TOON performed worse than both JSON and having the LLM just consume the human output.”
“The thing which is 98% JSON and absolutely didn't kill JSON?”
“that works remarkably well too”
“LLMs were less familiar with it, and thus spent even more tokens trying to decipher it”
“LLMs parse it correctly in our testing (GPT-4o, Claude, etc).”
“TOON separates schema from data—you declare field names once, then send pipe-delimited values.”
“Toon is 3 months old, it's not used by anyone, and it's therefore not in the training set of any model.”
“Is there evidence that LLMs adhere to this format better than to JSON? I doubt that.”
“Nice, it would be good idea to develop CFG for this as well”
“Yes, that's the purpose of TOON.”
“It’s the same story we’ve seen before with TOON.”
“TOON format: 50-58% token reduction for LLM interactions”
“TOON is a real game changer when used as an intermediary between the IDE/chat and the LLM”
“TOON’s advertised 40–60% savings dropped to ~2% overall once JSON was fully optimized.”
“TOON's structural overhead is lower than compact JSON for the structure families analyzed here”
“A free privacy-first, in-browser tool to instantly convert JSON to TOON”
“it actually makes a compelling use case for yet another format”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.