Hacker News: ac29

New comment by ac29 in "Pion, an agent designed to run any company autonomously"

ac29 — Mon, 14 Sep 2026 17:38:33 +0000

Relevant article about Andon Market: https://www.sfgate.com/local/article/san-francisco-market-ai...

New comment by ac29 in "PCB is brought to you by Fable 5"

ac29 — Mon, 14 Sep 2026 17:28:07 +0000

> I always wanted to try atopile - that might be a good match for AI/PCB work.

Unfortunately all public open source work halted 6 months ago and they pivoted to a browser based IDE product to monetize: https://atopile.io/blog/atopile-v16

New comment by ac29 in "PCB is brought to you by Fable 5"

ac29 — Mon, 14 Sep 2026 15:39:33 +0000

I sent in a larger, slightly more complex board to JLCPCB over the weekend and it was $175 for 5 boards with components and assembly. Shipping, tariffs (ouch), and sales taxes added another ~$100 or so.

New comment by ac29 in "Nvidia is the central bank of AI"

ac29 — Sat, 12 Sep 2026 22:46:12 +0000

gpt-oss-120b only has 5B active parameters, so its not surprising Qwen3.8 27B outperforms it (Qwen3.8 is also ~13 months newer, which is forever in LLMs)

New comment by ac29 in "Show HN: Copperhead – Cursor for circuit boards"

ac29 — Wed, 09 Sep 2026 01:53:52 +0000

I did that with Codex+Sol alone, and made something even more complex in less time with Codex+Astra over the weekend.

Its maybe more hands on than you want it to be, but you mostly just need to supervise the work. But yes, the CAD outputs go directly to the JLCPCB ordering form, or whereever you fab

New comment by ac29 in "Tiny $70 Xteink X3 e-reader"

ac29 — Mon, 07 Sep 2026 13:25:02 +0000

US prices dont include taxes, EU prices do

New comment by ac29 in "GPT-6 Astra on robot arms"

ac29 — Sun, 06 Sep 2026 02:59:44 +0000

GPT-6 is multimodal, LLMs alone have no vision capability

New comment by ac29 in "Dimethyl Mercury Exposure Incident at MIT"

ac29 — Mon, 31 Aug 2026 20:57:52 +0000

the methyl groups make it organic chemistry (and dangerous), elemental mercury is fairly safe

New comment by ac29 in "Apple caught off guard by AI demand for Mac Mini and Mac Studio"

ac29 — Mon, 31 Aug 2026 20:16:36 +0000

> AMD Strix Halo is about to release 192GB system which is a big upgrade to their current 128GB ai pc

Big upgrade to memory capacity but memory speed is only going up by a few percent, so its still going to be slow with more than a few B active params (I have one)

New comment by ac29 in "RISC-V is now officially supported by CPython"

ac29 — Sun, 30 Aug 2026 15:58:40 +0000

Micropython runs on tons of stuff and while it isnt capital-P Python, its close enough for doing a wide variety of embedded work without learning a new language/libraries/etc.

New comment by ac29 in "SiFive's First Server Platform"

ac29 — Tue, 25 Aug 2026 14:40:33 +0000

Because process affects performance per watt more than architecture so its a useful data point.

Even knowing nothing about performance, knowing that it is N6 means it probably isnt particularly power efficient (as compared to leading edge x86/ARM).

New comment by ac29 in "How Europe is killing makers and micro-entrepreneurs"

ac29 — Mon, 24 Aug 2026 14:51:14 +0000

> FCC allows suppliers to self-declare conformance with EMI/EMC rules

Yes, though they can also ask for your lab results or to inspect your testing facility and hit you with massive fines if it turns out you falsified your compliance documents. As you mentioned this is very rare though - even enforcement actions against non-certified intentional radiators are uncommon.

New comment by ac29 in "Kagi added a setting for removing paywalled links from search results"

ac29 — Fri, 21 Aug 2026 21:16:58 +0000

I have Kagi and ChatGPT subs and would agree ChatGPT is the better offering. But you can use Luna with Kagi Assistant on the cheap plan which works pretty well too

New comment by ac29 in "Codex on AWS bedrock bug causing 10x charges"

ac29 — Fri, 21 Aug 2026 04:48:07 +0000

There haven't been any free resets in the past week, there were 4 in the first half of the month

New comment by ac29 in "GLM-5.3 Artificial Analysis Benchmarks"

ac29 — Wed, 19 Aug 2026 01:23:00 +0000

Cost per task:

  Model                        Score    Cost / Task    Output Tokens / Task
  -------------------------------------------------------------------------
  Muse Spark 1.2 (xhigh)        56.8          $0.40                  30,430
  Gemini 3.7 Flash (high)       56.0          $0.40                  36,847
  GPT-5.6 Terra (max)           56.6          $0.51                  20,838
  GPT-5.6 Sol (high)            57.3          $0.52                   7,545
  GLM-5.2 (max)                 53.0          $0.56                  32,200
  GLM-5.3 (max)                 59.5          $0.68                  41,107
  GPT-5.5 (xhigh)               56.3          $0.69                  16,893
  Grok 4.6 (high)               60.9          $0.84                  21,735
  Kimi K3 (max)                 59.7          $0.84                  25,474
  GPT-5.6 Sol (xhigh)           59.0          $0.87                  11,098
  Qwen3.8 2.4T A95B             57.7          $0.95                  32,472
  Claude Opus 5 (medium)        58.6          $0.98                  12,459
  Qwen3.8 Max                   58.1          $1.13                  38,287
  GPT-5.6 Sol (max)             60.9          $1.23                  16,879
  Claude Opus 5 (high)          61.5          $1.52                  21,353
  Claude Opus 4.8 (max)         57.3          $1.65                  33,557

Benchmark score:

  Model                        Score    Cost / Task    Output Tokens / Task
  -------------------------------------------------------------------------
  Claude Opus 5 (high)          61.5          $1.52                  21,353
  GPT-5.6 Sol (max)             60.9          $1.23                  16,879
  Grok 4.6 (high)               60.9          $0.84                  21,735
  Kimi K3 (max)                 59.7          $0.84                  25,474
  GLM-5.3 (max)                 59.5          $0.68                  41,107
  GPT-5.6 Sol (xhigh)           59.0          $0.87                  11,098
  Claude Opus 5 (medium)        58.6          $0.98                  12,459
  Qwen3.8 Max                   58.1          $1.13                  38,287
  Qwen3.8 2.4T A95B             57.7          $0.95                  32,472
  Claude Opus 4.8 (max)         57.3          $1.65                  33,557
  GPT-5.6 Sol (high)            57.3          $0.52                   7,545
  Muse Spark 1.2 (xhigh)        56.8          $0.40                  30,430
  GPT-5.6 Terra (max)           56.6          $0.51                  20,838
  GPT-5.5 (xhigh)               56.3          $0.69                  16,893
  Gemini 3.7 Flash (high)       56.0          $0.40                  36,847
  GLM-5.2 (max)                 53.0          $0.56                  32,200

New comment by ac29 in "Gemini 3.7 Flash"

ac29 — Thu, 13 Aug 2026 21:04:40 +0000

> Using Google products in general is an effing nightmare as soon as you have to give them money

Spending money via Google Pay on Android is extremely easy, Google does know how to accept customer's money (in the consumer space)

New comment by ac29 in "Gemini 3.7 Flash"

ac29 — Thu, 13 Aug 2026 21:02:28 +0000

Last time I used the AI studio free tier it was limited to one or two requests, effectively useless. I think people are complaining its too hard to set up a paid API key (no reason you should make paying customers spend more than a few clicks and a minute of their time to pay you)

New comment by ac29 in "Grok 4.6"

ac29 — Thu, 13 Aug 2026 15:42:46 +0000

Muse Spark 1.2 benchmarks just shy of Opus/Sol and is significantly less expensive than Sonnet (which admittedly is overpriced). Haven't personally used it though

New comment by ac29 in "Grok 4.6"

ac29 — Thu, 13 Aug 2026 14:47:34 +0000

China has as least much engineering talent as the US, the claims that Chinese models must just be distilled US models feels like xenophobia.

There are more plausible explanations for why the models are similar - all the labs are buying the same datasets from third parties

New comment by ac29 in "Grok 4.6"

ac29 — Thu, 13 Aug 2026 13:39:57 +0000

I was under the impression labs released post-training checkpoints fairly often?

So Model N/N+1 might literally have had the exact the same pretraining run and only differ on how much/what kind of postraining they got