Hacker News: jakswa

New comment by jakswa in "Qwen 3.8 27B available on Cerebras at 1500 tokens/s"

jakswa — Thu, 03 Sep 2026 23:33:33 +0000

dang only for certain nvidia GPUs, had my hopes up

New comment by jakswa in "OpenAI begins rolling out GPT-6 Astra"

jakswa — Thu, 03 Sep 2026 20:55:35 +0000

don't see it in my AWS bedrock model list yet, but boy has bedrock mantle been annoying today with the errors/downtimes, with NO status page entries >_<

New comment by jakswa in "Show HN: We built open OpenRouter that turns usage into a better model"

jakswa — Fri, 28 Aug 2026 20:01:48 +0000

Oh. The UI screenshot on github is... not actually in the github repo? It's platform/hosted only? There's my first awkward discovery, but makes sense in retrospect.

New comment by jakswa in "Show HN: We built open OpenRouter that turns usage into a better model"

jakswa — Fri, 28 Aug 2026 13:20:22 +0000

I think I have to look at setting up one of these AI gateways for work, since AWS bedrock is such a PITA to hook a harness up to over IAM roles. Also I still can't believe bedrock hasn't released any open models in months (so there's paranoia that I'll want to swap in another provider).

Really though I'm hoping anthropic fixes the oppressive claude verbosity. I saw someone refer to being "clauderboarded" and my brain cannot let go of this as Claude's tokens bombard me.

New comment by jakswa in "Ornith-1.5: From Self-Scaffolding to Self-Improvement"

jakswa — Thu, 20 Aug 2026 18:26:00 +0000

ended up disabling ornith 9B. Oddly Ling 3 Tiny is pretty dang capable if its thinking is unleashed (tons of output tokens, maybe 5X the tokens but it's so fast it's maybe only twice as slow as a smarter model). This is a really interesting space if these super small models keep improving. SO FAST! Want to use!

New comment by jakswa in "Ornith-1.5: From Self-Scaffolding to Self-Improvement"

jakswa — Wed, 19 Aug 2026 17:50:50 +0000

I had to go down to UD-Q3_K_XL for Qwen 3.8 27B to get it to fit in VRAM and be usable, but I worry I'm gutting its intelligence somewhat. I too am interested in faster + more-usable alternative that can exchange blows with the Q3-dumbed 27B.

New comment by jakswa in "Ornith-1.5: From Self-Scaffolding to Self-Improvement"

jakswa — Wed, 19 Aug 2026 17:39:42 +0000

I'll be comparing the 9B vs Ling 3 Tiny (8B-A1B) as a scout model. Ling tiny is so fast but can be a little too dumb. Hope the 9B strikes a good middleground even if dense/slower.

New comment by jakswa in "Qwen3.8 27B scores 52 on Artificial Analysis"

jakswa — Mon, 17 Aug 2026 22:37:20 +0000

I've been waiting on this model to show up on the Deep SWE benchmark results and treat its absence/delay as an indication of how slow and unusable it is for good results. I bet it thinks to the moon on some of those complex challenges.

New comment by jakswa in "Qwen3.8 27B scores 52 on Artificial Analysis"

jakswa — Mon, 17 Aug 2026 21:39:48 +0000

Thanks for mentioning Ling 3 Tiny. This model has completely bypassed me and seems promising for how small it is.

New comment by jakswa in "Qwen3.8 27B scores 52 on Artificial Analysis"

jakswa — Mon, 17 Aug 2026 19:29:08 +0000

I'm waiting for this comparison too. I was impressed by a 1-shot GLM 5.3 did for me the other day.

New comment by jakswa in "Launch HN: Speko (YC S26) – OpenRouter for Voice AI"

jakswa — Mon, 17 Aug 2026 18:19:43 +0000

Gemma 4 (both E4B + 12B) performed really well as ears+brains. I mostly comment because I too am always scouting for a nice local all-in-one model.

New comment by jakswa in "Qwen 3.8 27B is excellent, but it defaults to overthinking things"

jakswa — Mon, 17 Aug 2026 02:55:55 +0000

I went back to Glimmer 30b for my 20GB of VRAM. Just a better experience fit-wise and speed-wise and tone-/voice-wise.

New comment by jakswa in "Gemini 3.7 Flash"

jakswa — Sat, 15 Aug 2026 23:13:27 +0000

thank you I'm still on ie8

New comment by jakswa in "Qwen 3.8 27B"

jakswa — Sat, 15 Aug 2026 00:20:53 +0000

I'm in the exact same boat with a 7900 XT and a good Glimmer 30B experience. I was really hoping qwen 3.8 would bring some memory/space efficiency savings along the lines of whatever is going on with Glimmer 30B. I have been surprised that a 30 billion model fits and runs better (at higher unsloth quantization! UD-Q4_K_XL fits!) than a 27 billion model.

New comment by jakswa in "Qwen 3.8 27B"

jakswa — Fri, 14 Aug 2026 18:26:11 +0000

anecdotes: 35B-A3B does want more memory, bigger model. But if you get it running it will be faster and more enjoyable to use -- text will fly by -- due to only 3B params being active, in my experience at least.

New comment by jakswa in "How Compaction Works in Pi"

jakswa — Fri, 14 Aug 2026 07:19:50 +0000

I dunno, I didn't read in-depth. Hopefully you don't gotta zoom in with human eyeballs.

New comment by jakswa in "How Compaction Works in Pi"

jakswa — Thu, 13 Aug 2026 22:56:46 +0000

OMP changed the default compaction to images! Kinda nuts to read about. Saves the generation cost of the traditional compaction step and writes the context as tiny text to an image, if I was following correctly.

New comment by jakswa in "Gemini 3.7 Flash"

jakswa — Thu, 13 Aug 2026 19:14:50 +0000

my whole world is shifting. have I been seeing _different pelicans_ from everyone else?!

New comment by jakswa in "Gemini 3.7 Flash"

jakswa — Thu, 13 Aug 2026 18:34:28 +0000

This pelican gave me a good laugh, because there's enough reasoning that the render is out of sight initially. The buildup!

New comment by jakswa in "Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows"

jakswa — Mon, 10 Aug 2026 16:58:37 +0000

I'll back up your smaller claim, but be specific that it's UD-Q4_K_XL size:

- muse glimmer: 15.9GB

- qwen 3.6 27B: 17.6GB

My video card is so close to its limit that these GB thresholds are mattering too much for me :D