96.0% on LongMemEval: what retrieval does for long-term memory
Comments on “96.0% on LongMemEval: what retrieval does for long-term memory”
3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked
Papr reports 96.0% accuracy (480 of 500) on the LongMemEval-S benchmark, the highest published score the company says it has found, and says the run used about 16.5k tokens of context per question at its top-5 setting.
Why it ranked
No. 55 of 60 with a score of 0.56. The panel rated it highest on new information (0.88) and substantiated (0.83), lowest on significance (0.54). The judges disagreed most on promotional (spread 0.48), so the story is marked contested. Placed No. 55, where the score alone would also have put it: the 22nd AI & machine learning story of the day (−0.080). The placement terms move stories; they are not part of the score.
- AI judgeJudge 2claude-haiku-5-5
- Benchmark claims for LLM memory retrieval with detailed methodology and public artifacts, but it is a vendor-driven results post with tuned configs and a small variance margin, so promotional and overclaim risk are moderate.
- AI judgeJudge 3gpt-5.6-luna
- A reproducible engineering evaluation of long-term memory retrieval, with benchmark data, pipeline details, latency and context-cost tradeoffs. It is highly actionable, though self-reported, tuned, and partly promotional because the system and embedding API are commercial.
Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.
Score
0.56total under rubric v2.3Contestedjudges differ by 0.48 on promotional
- Significance0.54
- Reader interest0.71
- Audience breadth0.55
- Practical relevance0.75
- Substantiated0.83
- Technical depth0.77
- New information0.88
- Counts against: Promotional0.42
- Counts against: Overclaim0.36
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| SignificanceHow much does this change what people in tech know or can do? | 0.56 | 0.40 | 0.67 | 0.54 | 0.27 | +0.24 | +0.130 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.66 | 0.55 | 0.93 | 0.71 | 0.38 | +0.26 | +0.185 |
| Audience breadthHow many people who build technology does it concern? | 0.53 | 0.45 | 0.67 | 0.55 | 0.22 | +0.13 | +0.071 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.84 | 0.60 | 0.82 | 0.75 | 0.24 | +0.22 | +0.166 |
| SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations. | 0.96 | 0.70 | 0.84 | 0.83 | 0.26 | +0.05 | +0.042 |
| Technical depthHow technically substantive is it? | 0.88 | 0.65 | 0.78 | 0.77 | 0.23 | +0.05 | +0.038 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.87 | 0.80 | 0.96 | 0.88 | 0.16 | +0.05 | +0.044 |
| PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional. | 0.16 | 0.45 | 0.64 | 0.42 | 0.48 , above the contested threshold | −0.19 | −0.079 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.31 | 0.40 | 0.38 | 0.36 | 0.09 | −0.11 | −0.040 |
| Sum of contributions | +0.558 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.558 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-11-16/55/"><img src="https://rubric.hrih.si/badge/2026-11-16/55.svg" alt="Ranked #55 on Rubric, 16 Nov 2026" width="245" height="28"></a>Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 1 pts, not on front page
Recorded for context only, checked 16 Nov 2026, 05:47 UTC. It played no part in the ranking.