Infinity: A runtime for extreme agent concurrency
Comments on “Infinity: A runtime for extreme agent concurrency”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
AWS and UC Berkeley released Infinity, an Apache 2.0 research runtime from the Hydro project that treats each agent tool call as a suspension point, like a stackless coroutine. It saves the conversation and ends the process, then resumes when the result arrives on a queue. In AWS's single-thread benchmark with a simulated model, idle agents used about 100 KB each, so roughly 75,000 fit in a Raspberry Pi's 8 GB. The reference deployment uses SQS, Lambda, Aurora DSQL and DynamoDB. The figures are self-reported and unverified.
Why it ranked
No. 16 of 93 with a score of 0.65. The panel rated it highest on new information (0.92) and technical depth (0.88), lowest on audience breadth (0.57).
Every answer behind this is in the table under Score.
Score
0.65total under rubric v2.2
- Significance0.60
- Reader interest0.78
- Audience breadth0.57
- Practical relevance0.70
- Substantiated0.75
- Technical depth0.88
- New information0.92
- Counts against: Promotional0.13
- Counts against: Overclaim0.27
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) |
|---|---|---|---|---|---|---|---|
| Significance | 0.62 | 0.45 | 0.72 | 0.60 | 0.27 | +0.24 | +0.143 |
| Reader interest | 0.73 | 0.72 | 0.90 | 0.78 | 0.18 | +0.26 | +0.204 |
| Audience breadth | 0.49 | 0.60 | 0.62 | 0.57 | 0.13 | +0.13 | +0.074 |
| Practical relevance | 0.73 | 0.60 | 0.78 | 0.70 | 0.18 | +0.22 | +0.155 |
| Substantiated | 0.87 | 0.55 | 0.82 | 0.75 | 0.32 | +0.05 | +0.037 |
| Technical depth | 0.94 | 0.78 | 0.91 | 0.88 | 0.17 | +0.05 | +0.044 |
| New information | 0.95 | 0.85 | 0.96 | 0.92 | 0.11 | +0.05 | +0.046 |
| Promotional | 0.08 | 0.20 | 0.12 | 0.13 | 0.12 | −0.19 | −0.025 |
| Overclaim | 0.32 | 0.30 | 0.18 | 0.27 | 0.14 | −0.11 | −0.029 |
| Sum of contributions | +0.649 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.649 | ||||||
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/2026-10-25/16/"><img src="https://rubric.hrih.si/badge/2026-10-25/16.svg" alt="Ranked #16 on Rubric, 25 Oct 2026" width="245" height="28"></a>Claim check
Claim check: 8 no source given. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
No source givenVersion
Infinity is released under the Apache 2.0 open source license as part of the Hydro project, a collaboration between AWS and UC Berkeley.
The article gives no link or citation for it.
The article says “we’re releasing Infinity under the Apache 2.0 open source license as part of the Hydro project, a collaborative effort between AWS and UC Berkeley”
The source says No source to compare with.
No source givenNumber
About 75,000 concurrent Infinity agents fit in the 8 GB of memory of a Raspberry Pi 5.
The article gives no link or citation for it.
The article says “we can fit 75,000 concurrent agents in the 8 GB of a Raspberry Pi 5”
The source says No source to compare with.
No source givenBenchmark result
In the benchmark, Infinity was run with 80,000 agents (3.2 million completions) on a single thread of an AMD EPYC 9R14 with in-memory stores.
The article gives no link or citation for it.
The article says “We ran the benchmark with 80,000 agents, for a total of 3.2 million completions, on a single thread of an AMD EPYC 9R14 with in-memory stores”
The source says No source to compare with.
No source givenNumber
Resident memory in the benchmark grew linearly at about 100 KB per idle agent.
The article gives no link or citation for it.
The article says “Resident memory grew linearly, at about 100 KB per idle agent”
The source says No source to compare with.
No source givenCausal conclusion
Most of the per-agent memory is taken up by conversation histories, since an idle agent has no task, stack, or connection.
The article gives no link or citation for it.
The article says “Most of this memory is taken up by the conversation histories, since an idle agent has no task, stack, or connection associated with it”
The source says No source to compare with.
No source givenCausal conclusion
When Infinity suspends an agent during a tool call, the serverless function can end, so even a multi-day tool call needs nothing running while the agent waits.
The article gives no link or citation for it.
The article says “When Infinity suspends an agent during a tool call, the serverless function can end. So even if a tool call takes days, nothing needs to run while the agent waits”
The source says No source to compare with.
No source givenVersion
The reference AWS deployment uses an SQS FIFO queue with one MessageGroupId per thread, Lambda functions running one step each, Aurora DSQL for conversation histories and DynamoDB for deduplication state.
The article gives no link or citation for it.
The article says “Our reference deployment on AWS uses an SQS FIFO queue as the input queue, with one MessageGroupId per thread”
The source says No source to compare with.
No source givenGeneralization
Appending a second result to a tool call violates the one-result rule, and models typically reject such a conversation.
The article gives no link or citation for it.
The article says “Appending a second result to the original tool call violates the one-result rule, and models typically reject such a conversation”
The source says No source to compare with.
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 25 Oct 2026, 05:46 UTC. It played no part in the ranking.