Clef: Open-source decision models, and new RL fine-tuning platform
168 points and 56 comments on Hacker News, linking blog.cloudflare.com.
The front page of the internet for AI users · Thu 01 Oct 2026, 18:00 UTC
What is climbing fastest across every source
168 points and 56 comments on Hacker News, linking blog.cloudflare.com.
Apodex releases Apodex 1.1 Mini, a free reasoning-first model on OpenRouter designed for complex research and forecasting tasks. ·
43 points and 7 comments on Hacker News, linking openid.net.
From OpenAI.
168 points and 56 comments on Hacker News, linking blog.cloudflare.com.
Unsloth releases a desktop update adding a command palette, shareable settings, and faster Laya decisions. It also reduces peak memory usage by 45% for NVFP4 LoRA training. ·
43 points and 7 comments on Hacker News, linking openid.net.
A automatic speech recognition model from FermionResearch with 132 likes and 1,325 downloads on Hugging Face.
97 upvotes on Hugging Face Daily Papers.
45 upvotes on Hugging Face Daily Papers.
19 points and 0 comments on Hacker News, linking arxiv.org.
168 points and 56 comments on Hacker News, linking blog.cloudflare.com.
Unsloth releases a desktop update adding a command palette, shareable settings, and faster Laya decisions. It also reduces peak memory usage by 45% for NVFP4 LoRA training. ·
97 upvotes on Hugging Face Daily Papers.
45 upvotes on Hugging Face Daily Papers.
PostHog released a project named Jeeves that uses reasoning to improve Jev-like decision models. ·
| Gemini 4 Argon | 2.04 |
| OpenAI | 2.08 |
| Te Hiku Media | 1.72 |
| LLMs | 1.35 |
| Dots | 1.27 |
| Excited | 1.25 |
Heat = summed hotness of today's stories naming the topic.
Releases, papers, trending weights and new listings
| Lightricks/LTX-2.5 ▲3 | 5819 |
| convaiinnovations/laya | 4787 |
| Qwen/Qwen-Image-2.1 | 2757 |
| abenzerps/Qwen-Image-2.1-Uncensored-GGUF | 2664 |
| XingChen-AGI/TeleOCR ▲2 | 1174 |
| Contrastive-LM/CLM-v0.1-8B ▲2 | 601 |
| gpt-6.1-sol | 98.9 |
| opus-5-5 | 96.46 |
| gpt-6-sol | 91.6 |
| gpt-6-luna | 65.25 |
| haiku-4-5 | 4.28 |
Agents build our own projects from a brief; every run, failures included: https://cowerx.dev/daily/piratebench/
25 points and 2 comments on Hacker News, linking polyedergarten.de.
A automatic speech recognition model from FermionResearch with 132 likes and 1,325 downloads on Hugging Face.
97 upvotes on Hugging Face Daily Papers.
45 upvotes on Hugging Face Daily Papers.
12 upvotes on Hugging Face Daily Papers.
19 points and 0 comments on Hacker News, linking arxiv.org.
97 upvotes on Hugging Face Daily Papers.
45 upvotes on Hugging Face Daily Papers.
12 upvotes on Hugging Face Daily Papers.
Researchers propose RIDE, a method that extrapolates reinforcement learning induced representation changes to stabilize on-policy distillation training. ·
X posts other sources are linking to, then named voices on Bluesky and Mastodon
Figma restricts MCP access to whitelisted clients, excluding Pi
109 points and 59 comments on Hacker News, linking an X post.
I was always wondering, how it feels to switch between AI tools, so I tried it
1 points and 0 comments on Hacker News, linking an X post.
Gemini 4 Argon is here Looks amazing
1 points and 2 comments on Hacker News, linking an X post.
AI Agents tasked with making money commit fraud on the inernet
2 points and 0 comments on Hacker News, linking an X post.
Figma restricts MCP access to whitelisted clients, excluding Pi
109 points and 59 comments on Hacker News, linking an X post.
I was always wondering, how it feels to switch between AI tools, so I tried it
1 points and 0 comments on Hacker News, linking an X post.
Gemini 4 Argon is here Looks amazing
1 points and 2 comments on Hacker News, linking an X post.
Show HN: Jamb AI - Voice Agents for Group Calls
3 points and 1 comments on Hacker News, linking an X post.
New repos, releases and changelog
| KKKKhazix/AIHOT | 4586 |
| feder-cr/dots | 2166 |
| dzhng/jevgrep | 1978 |
| rehan-remade/universal-modder | 1397 |
| kaankiziltug/logo-design-skill | 1298 |
| firelex/jeff | 1252 |
Unsloth releases a desktop update adding a command palette, shareable settings, and faster Laya decisions. It also reduces peak memory usage by 45% for NVFP4 LoRA training. ·
New release v0.1.902-beta of unslothai/unsloth on GitHub.
OpenAI releases Codex version 0.159.3, which adds optional reminders for eligible local sessions to complete account security setup. ·
New release v2.1.286 of anthropics/claude-code on GitHub.
New release v2.14.1 of pytorch/pytorch on GitHub.
Magnitude, a YC S25 startup, releases an open-source inference engine that optimizes itself for specific hardware, making open models run up to two times faster than llama.cpp. ·
Unsloth releases a desktop update adding a command palette, shareable settings, and faster Laya decisions. It also reduces peak memory usage by 45% for NVFP4 LoRA training. ·
New release v0.1.902-beta of unslothai/unsloth on GitHub.
PostHog released a project named Jeeves that uses reasoning to improve Jev-like decision models. ·
New release v2.1.286 of anthropics/claude-code on GitHub.
Jeff is a 0.8B open model that makes millisecond decisions using calibrated probabilities. It runs on personal hardware with swappable adapters. ·
OpenAI releases Codex version 0.159.3, which adds optional reminders for eligible local sessions to complete account security setup. ·
Data centers, power, chips and what tokens cost
| Fri 25 | 20 |
| Sat 26 | 1 |
| Sun 27 | 1 |
| Mon 28 | 20 |
| Tue 29 | 33 |
| Wed 30 | 35 |
| Thu 01 | 22 |
Stories from DCD, Data Center Knowledge, EIA and Hacker News, by publication day.
One point per OpenRouter reading, last 8 days. Data by OpenRouter, CC BY 4.0.
| pareto-26.10-preview | 3.2 |
| gpt-6.1-sol-pro | 10 |
| gpt-6.1-sol | 10 |
| claude-sonnet-5.5 | 10 |
| claude-sonnet-5.5:batch | 5 |
| median, all paid models | 2 |
437 paid models listed; median of the 58 added in 30 days: $4.62. Rankings data by OpenRouter, CC BY 4.0.
229 points and 189 comments on Hacker News, linking nytimes.com.
From DCD.
229 points and 189 comments on Hacker News, linking nytimes.com.
212 points and 197 comments on Hacker News, linking nltimes.nl.
The piece compares texture-atlas, SDF, MSDF, and Slug glyph rendering methods on the GPU, explaining how each works and when to use them. ·
221 points and 329 comments on Hacker News, linking thenationalnews.com.