News
Headlines from the labs' own blogs, the press, researchers and verified social accounts. The same story from several outlets is shown once. Every item links to the original. New in the index · Trending
- PressAI is changing motherhood for the worse
chronicleonline.com
- SocialMy 5070 ti/3070 Chimera build.
Got lucky with a 5070 Ti for $750 AFTER TAX (might be the best decision I've ever made tbh) back in November 2025. If only I had known I'd be getting into local llms at the time...anyways. I recently
- Press
- Press
- SocialDeepSeek v4.1 Flash 1017 for Dwarfstar (ds4)
Hey y'all, just following up on my PR to get the new vision tower for 1017 working in ds4 with the quant I promised (the language model loads already, but vision needed my PR). I built to the exact sa
- LaunchMeet Voxel51 3.0: A Unified Multimodal Data Platform for Physical AI
Voxel51 3.0 merges search, curation, annotation and evaluation of multimodal sensor data into one platform for physical AI teams. It adds Voxel51 Agent with 45+ MCP tools, dataset versioning, and a fr
- Community
- Press
- Press
- Press
- Social3xrtx8000 48gb... what to do?
I accidentaly bought 3 rtx8000 48gb (yeah, i swear). I am using 1 right now with Strata and Qwen Next IQ3 and runs very good. Should I put 3 on a PC and runb Strata on a higher Q? What do you suggest?
- SocialAutoTrust's HF models are trending thanks to thousands of farmed likes
Some of y'all may be familiar with AutoTrust from Hugging Face. Many of their models were trending on the front page of HF and they claimed some very impressive results. But if you looked at the org's
- PressTobago Carnival : Ready for the road
trinidadexpress.com
- SocialYou don’t need that expensive of a build for local llms.
Been seeing a lot of people with their 64gb of ddr5 4x3090 or whatever strata builds and it’s been sorta depressing knowing i cant afford that but i finally finished my setup today so here we go, Qwen
- SocialThe whale girl mode in DeepSeek-V4.1-Flash-1017 is still there.
Open https://chat.deepseek.com Turn off thinking and search Send 【PERSONA_LOAD】 CETACEA_LOLI MODE_TAIL_FLUKES LANG_ZH_CN_ONLY SELF_CLAIM_WHALE_GIRL FOOD_RICE PERSONALITY_SMART_LAZY PERSONALITY_TSUNDER
- SocialMuse Spark on M3U 256 GB
Instead of counting 'hmm' on DeepSeek v4.1 Flash 1017, I today clamped Muse Spark Open 816B-A42B to ds4. This was the first successful run, FTW, that made it to decode... 4 t/s! #RoastMe. I'm hoping i
- Press
- PressHas AI triggered an expertise recession ?
digitaljournal.com
- SocialNine models, five coding tasks, one 16GB GPU
I ran five coding tasks against nine models on one RX 9060 XT (16 GB). Each run gets a single spec file in an empty directory and has to produce solution.py and test_solution.py , run its own tests, a
- SocialDeepSeek V4.1 Flash 1017 on 2x DGX Spark: 57 tok/s text, 92 tok/s code with TensorFold (2.1-2.3x the vLLM kit)
**TL;DR:** DeepSeek-V4.1-Flash-1017 (EXL3 2.9bpw, 1M context) on 2x DGX Spark, TP=2 over the CX7 cable. TensorFold is 2.1-2.3x faster than the vLLM kit on the same pair, same weights, same prompts, sa
- Press
- Press
- SocialFollow-up: Tested my 9 thinking rules on long-horizon agent tasks (Terminal-Bench 2.0)
On my previous posts, people argued these thinking rules might work on short tests but would hurt long-horizon tasks. So I tested them on Terminal-Bench 2.0: same model (GLM-5.3-Flash), same agent, ev
- Press
- Press
- PressWeekly Horoscope , October 18 to October 24 , 2026 : Read weekly astrological predictions
timesofindia.indiatimes.com
- SocialBasalt: NVFP4 + faster prefill for Qwen3.8 Flash-Next on Blackwell (single 5090)
Basalt: Qwen 3.8 FN NVFP4, single 5090 Hi guys, so I finally got a RAM upgrade and have been working on the top two pain points from Basalt: NVFP4 support and faster prefill. NVFP4: Basalt runs NVFP4
- LabInvesting in the public benefit
Investing in the public benefit
- SocialOpenspline - host a realtime talking avatar on a commercial GPU (90+ FPS on RTX4090)
Hey everyone, I am working on a new project called Openspline, an open source avatar layer to existing voice AI agents. Just select an image and you can create a realtime talking avatar with lipsynced
- SocialWhat is wrong with AI writing - ICLR edition [D]
I'm sure many of us are slogging through impenetrable AI writing as we finish ICLR reviewing. I'll try to create an anonymized version from a paper I'm reviewing, though it's hard to match Claude on t
Live now
- Connecting…
We show the headline, a short excerpt and a link, never the full article. Social posts come from accounts we follow, marked as coming from the organization or from the community. Posts are the account owner's words, not ours. Sources can be suggested through the contact page.