44za12
Re:Hybrid Reasoning Agents: LLM agents that think in latent space, act...
on Sunday October 18, 2026 @05:11PM (#50145525)
Where are the benchmarks? Does it just speed up inference and reduce token usage or does it seem to improve accuracy on certain tasks as well?
Re:Choosing an AI model: one prompt, 11 models, different results
on Thursday August 13, 2026 @04:59PM (#49291780)
Built rightsize exactly for this. https://nehmeailabs.com/right-size
Re:Show HN: Segue – Save context in one AI, load it in another by a sh...
on Tuesday July 28, 2026 @11:46AM (#49085670)
Others have already pointed out the absurdity, but just to put it across. Why not just ask first agent to update AGENTS.md adding context and next steps, and ask the next agent to continue from there?
Re:Nativ: Run frontier open models locally on your Mac
on Monday July 20, 2026 @03:36PM (#48983821)
+1 came here to say this, I opened the link expecting some technical breakthrough. Misleading click bait title.
Re:Ask HN: Who wants to be hired? (July 2026)
on Wednesday July 01, 2026 @03:15PM (#48751801)
Location: UAE
Remote: Preferred
Want to relocate: No
Philosophy: Brutally Efficient
More: https://aazar.me
Horcrux – Distributed, Zero-Trust Secret Manager
Lattice Deduction Transformers
MiniMax M3
Δ-Mem: Efficient Online Memory for Large Language Models
Beyond Semantic Similarity
Re:I run multiple $10K MRR companies on a $20/month tech stack
on Sunday April 12, 2026 @04:18AM (#47737256)
I read it as an article in defence of boring tech with a fancier/clickbaity title.
Here’s the more honest one i wrote a while back:
Re:Qwen3.5 Fine-Tuning Guide – Unsloth Documentation
on Wednesday March 04, 2026 @11:47AM (#47250200)
Specialised models easily beat SOTA, case in point: https://nehmeailabs.com/flashcheck
Re:Number Research Inc
on Wednesday March 04, 2026 @12:14AM (#47243376)
All of us use the same keyboards more or less, maybe us randomly typing a large number is not as random as we would like to think. Just like how “asdf”, “xcyb” are common strings because these keys are together, there has to be some pattern here as well.
In Defense of Boring Technology
Show HN: RightSize CLI, Find the cheapest LLM that works for your prompt
https://github.com/NehmeAILabs/rightsize-cli
RightSize runs your prompt against candidate models (Kimi, GLM, Qwen, Gemma etc.) in parallel via OpenRouter. Then it uses a stronger model as a Judge to score accuracy against a baseline model.Happy to answer questions.
Re:I cut Claude API costs from $70/month to pennies
on Monday January 26, 2026 @08:24AM (#46765335)
Yes, I included a 'Model Selection Cheat Sheet' in the README (scroll down a bit).
I map them by task type:
Tiny (<3B): Gemma 3 1B (could try 4B as well), Phi-4-mini (Good for classification). Small (8B-17B): Qwen 3 8B, Llama 4 Scout (Good for RAG/Extraction). Frontier: GPT-5, Llama 4 Maverick, GLM, Kimi
Is that what you meant?
Show HN: LLM Sanity Checks – A practical guide to not over-engineering AI
https://github.com/NehmeAILabs/llm-sanity-checks
I keep seeing teams use frontier models for tasks a regex or a 4B model could do cheaper and faster.This repo is a collection of opinionated patterns and heuristics to help you rethink the architecture of your AI workflows and ease the decision-making process while ensuring maximum efficiency.It covers:- A decision tree for architectural sanity checks. - Tradeoffs between JSON and...
Re:I cut Claude API costs from $70/month to pennies
on Monday January 26, 2026 @02:03AM (#46762668)
This is the way. I actually mapped out the decision tree for this exact process and more here:
Re:Stop using JSON for LLM structured output
on Saturday January 24, 2026 @02:46AM (#46741820)
For simple extraction tasks, a delimiter-separated string uses 11 tokens vs 35 for JSON. Output tokens are the latency bottleneck.
Stop using JSON for LLM structured output
FlashCheck-270M: Open weights for fact verification (Apache 2.0, WASM Demo)
Re:De-Googling TOTP Authenticator Codes
on Monday September 01, 2025 @08:35AM (#45092249)
Love the minimalism.
Re:De-Googling TOTP Authenticator Codes
on Monday September 01, 2025 @06:21AM (#45091377)
Shameless plug.
I’ve been using a cli tool i had created for over 2 years now, it just works. I had more ideas but never got to incorporate those.
Re:Free AI Tool Removes Image Backgrounds in Seconds
on Saturday August 30, 2025 @03:16AM (#45072582)
Have been using remove.bg for this for years now.
Re:Show HN: Rust macro utility for batching expensive async operations
on Sunday August 17, 2025 @11:03AM (#44932116)
Like a sempahore?
Re:Gemma 3 270M: The compact model for hyper-efficient AI
on Thursday August 14, 2025 @12:55PM (#44902789)
I’ve had great luck with all gemma 3 variants, on certain tasks it the 27B quantized version has worked as well as 2.5 flash. Can’t wait to get my hands dirty with this one.
Re:Benchmarking GPT-5 on 400 real-world code reviews
on Friday August 08, 2025 @02:20AM (#44834069)
Can you benchmark Kimi K2 and GLM 4.5 as well? Would be interesting to see where they land.
That was quick, vibe coded, I presume?
The ability to submit a story using a curl would have been fun.