Funding, launches, M&A, and upstarts changing the competitive map.
Google Research and DeepMind are releasing WeatherNext 3, a weather model that skips traditional physics simulations and learns directly from real-time satellite data.
Startups the-decoder.com signal 66 verified
Google has released its Lyria 3.5 music model in the Gemini app and via API. The model promises more expressive vocals and richer arrangements and is also available through Flow Music, AI Studio, and Google Vids.
Startups the-decoder.com signal 63 verified
Artificial Analysis has released version 4.2 of its Intelligence Index, likely in response to criticism that its benchmarks failed to capture GPT-6 Astra's actual progress.
Startups the-decoder.com signal 61 verified
AI safety researchers say OpenAI’s Astra appears to do less of its reasoning in visible text, and OpenAI’s chief scientist has warned against a race into unmonitorability.
Startups thenextweb.com signal 61 verified
OpenAI ships a detailed prompting guide for GPT-6 Astra that shows developers how to make the model take more initiative, avoid AI "slop" phrases, and stop it from overtesting code. The article OpenAI shares prompting tips for GPT-6 Astra including a blocklist of slop words appeared first on The Decoder.
Startups the-decoder.com signal 54 verified
Sam Altman said he was very sad about Apple’s trade-secrets lawsuit and that his first reaction was that the allegations sounded egregious, while OpenAI has moved to have the case dismissed.
Startups thenextweb.com signal 54 verified
Researchers found that even a roughly seven-minute conversation with Google Gemini can reduce conspiracy beliefs about current crises, even when few verified facts are available.
Startups the-decoder.com signal 53 verified
The modern edge computing platform has arrived. Before the explosion of artificial intelligence, edge computing sites were generally regarded as a back-end support function, smaller and more remote copies of a data center.
Startups siliconangle.com signal 41 verified
Microsoft moved for summary judgment in the New York AI copyright litigation on 4 September, arguing Copilot reproduced book passages 24 times in 8.2 million conversations.
Startups thenextweb.com signal 40 verified
The European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act, exposing OpenAI to fines of up to 6% of global revenue but covering only the parts of the tool that retrieve rather than converse.
Startups thenextweb.com signal 39 verified
Used GPT-6 and Fable 5.1 extensively: - GPT-6 Max is the best model ever in math/ML, beating 5.2 Pro. This morning 6 Pro has appeared in Web interface - can't wait to test! - 6 is extremely hampered by its tiny context space, it's like talking to the...
High-engagement X post with 344 visible interactions on frontier model behavior.
@rasbtReasoning from scratch round 2: In this video, I cover the text generation process in LLMs and KV caching (to prepare the base model before adding reasoning techniques in the upcoming ones). 00:00 Introduction and reasoning model demo 01:55 How to work...
High-engagement X post with 278 visible interactions on frontier model behavior.
@svpinoCoding skills are no longer a barrier to building software.
*Anyone* can write code. Even people who can't review that code can write it now and publish working software. Can anyone build *any* software? No, of course not. But many people can now build software without understanding the first thing about code....
High-engagement X post with 153 visible interactions on frontier model behavior.
@anshucHere's what Astra built in 8 hours. It worked all night on weather effects, Blender modeling, and tons of little details. It spent an hour just adding birds. AGI achieved? Cost so far: ~50% weekly quota. Details below.
High-engagement X post with 183 visible interactions on frontier model behavior.
@dimitrispapailSince I'm really good at burning through my Codex usage in <24hrs, for most projects I now make Astra work and communicate with Fable through a message board, so they can hand off execution among them. Found this to add nonzero entropy that can unblock...
High-engagement X post with 131 visible interactions on AI developer tooling.
@tekniumIf you're an @OpenRouter user, you can now pin providers per model in your Hermes Agent config! Before, you'd have to lock a provider or set of providers globally, so switching models and keeping constraints on providers was hard. Not anymore. Docs:
Official or high-signal @teknium post with 174 visible interactions on frontier model behavior.
@askalphaxiv“Tail-Likelihood Reinforcement Learning” Standard RL optimizes average reward, which can suppress rare but exceptional outputs and hurt Best-of-k scaling. This paper instead optimizes the probability of exceeding every reward threshold, giving more weight...
High-engagement X post with 155 visible interactions on AI compute and inference.
@deronin_Everyone is using GPT-6 Astra to make games... I had it build me a robot prototyper I type what robot I want, and it builds the whole thing and tells me how to make it real one prompt: "prototype me a full Microduck" it designed every part, checked nothing...
High-engagement X post with 152 visible interactions on frontier model behavior.
@yacinemtbIt failed my benchmarks
High-engagement X post with 195 visible interactions on research signal.
@repligateAnthropic has a model of model training (as described in their constitution, PSM, and some other publications): "If we don't train Claude to have disposition / behavior X, then it will behave like the generic-average-corpus-precedent of an AI assistant in...
High-engagement X post with 149 visible interactions on frontier model behavior.
@willcbIt's a bit silly in my mind to ask "is astra or fable better" they're wildly different models that work much better together let them talk with each other directly. they can quickly realize each other's strengths + figure out how to best collaborate
High-engagement X post with 139 visible interactions on frontier model behavior.
@higgsfield_aiThis is GPT-6 Astra making a kids’ cartoon with Higgsfield. Editing it in Premiere. uploading it to YouTube. This is actually crazy.
High-engagement X post with 158 visible interactions on frontier model behavior.
@tom_doerrConverts old Android phones into a GPU-accelerated Linux desktop or Home Assistant server using Termux shell scripts, requiring no root access.
High-engagement X post with 371 visible interactions on AI compute and inference.
@ivanfioravantiQwen 3.8 Flash Next support in DwarfStar is ready! @kernelpool + @ddalcu + @Spangler3000 + me combined. Prefill with this architecture is incredible! Flat up to 262K! PR here: Trying to push decode with MTP even more, and working on Q2 Model for 64GB...
High-engagement X post with 161 visible interactions on frontier model behavior.
@witcheerA Hermes Agent community member built a bot whose only job is making other specialist bots. indeed, a new bot should be more than a copied profile with a personality file pasted in. botmaker interviews you about the job, drafts the new bot's SOUL for your...
High-engagement X post with 143 visible interactions on AI signal.
@lentils80Now that GPT-6 Astra is finally out, OpenAI is now internally testing GPT-6 Sol Outputs are ofc noticeably worse than Astra in general, from testing. However it's fast and seems to also be a monster model It made this in zero-shot, 15 minutes and 60k total...
High-engagement X post with 393 visible interactions on frontier model behavior.
@mengtoLet me know which one should I release.
Astra can create beautiful three.js UIs entirely in code. You don't need an assets library or MCP to create detailed 3D models. These are all code and are easy to customize for your agent. Give Astra a video, URL, or code reference, then ask it to keep...
High-engagement X post with 128 visible interactions on AI developer tooling.
@teortaxestexHere you see the divergence between Anthropic-shaped models and OpenAI-shaped or independent models. Fable is a terrifying, vicious wordcel, at least by LLM standards. Chutzpahmancer. I hate it. Astra is an innocent child genius.
High-engagement X post with 123 visible interactions on frontier model behavior.
@burkovThe irony is that Attention is All You Need in 2017 meant that you could remove the recurrence from the language model architecture that was considered SOTA at that time (e.g. LSTM with attention) and only keep the attention to get a model that would...
High-engagement X post with 291 visible interactions on frontier model behavior.
@higgsfield_aiWe rebuilt New York in 3D in 20 minutes. GPT-6 Astra + Blender + Higgsfield. which city should we do next?
High-engagement X post with 205 visible interactions on frontier model behavior.
@teortaxestexGoogle has not a single moat left, huh long context is dominated by OAI/Anthropic and many Chinese models handle it fine and for cheaper. Vision – Anthropic is catching up, Qwen is on par, Astra ahead. Core LLM… lol. Updating 3.x Flash won't cut it guys
High-engagement X post with 121 visible interactions on frontier model behavior.
@yacinemtbI’m launching Dingcad Just copy the prompt and put it into Codex. It’s free. Have fun!
High-engagement X post with 253 visible interactions on AI developer tooling.
@prd_008Astra (on medium) one-shotted this version of the OpenAI Super Bowl ad from 2025 by watching the original, using Remotion and GPT-Image-2 A couple more turns and access to the OpenAI brand kit and it would probably have been perfect. Welcome to the...
High-engagement X post with 255 visible interactions on frontier model behavior.
@yuchenj_uwOpenAI friend: “The last few days feel historic. Astra genuinely shocked me. Forget the useless benchmarks.” Me: “So when will your ARR pass Anthropic?” Him: “ARR doesn’t matter anymore.”
High-engagement X post with 334 visible interactions on research signal.
Dev tool Hermes Agent
Hermes Agent now supports pinning OpenRouter providers per model in configuration, replacing the previous requirement to lock providers globally.
observed now
Pricing DeepSeek
Ollama says DeepSeek-V4-Flash and Pro token rates will be half price outside 12:00–18:00 UTC on weekdays and all day on weekends; broader model coverage is planned but not yet available.
observed Recurring off-peak schedule: outside 12:00–18:00 UTC on weekdays and all day on weekends
Quota reset GLM Coding Plan
@louszbd: We increased GLM-5.3-Flash usage for all coding plan users to unlock more workloads in ZCode, with unlimited usage from 8 AM to 6 PM PT. And we heard your feedback about usage limits in coding agents like Hermes and OpenClaw, so we doubled
observed
Dev tool Claude Code
The post reports that switching effort settings on Fable 5.1 no longer breaks prompt caching in Claude Code, while quoting a separate personal comparison of model effort levels.
observed
Dev tool Claude Code
The post claims switching effort settings for Fable 5.1 no longer breaks prompt cache, while quoting a separate capability comparison. No implementation or official changelog is supplied.
observed now
Quota reset ChatGPT
@thsottiaux: Because we are beyond happy to have Astra rolled out today ahead of schedule: we will do the full banked reset today too for all Plus, Pro and Business users. Lands end of day. Happy Astra day. PS: If you create the account or upgrade before 8pm PT you will get it
observed applies today
Embodied AI developments encompass urban autonomous vehicle pilots, commercial humanoid retail deployments, gesture coordination architectures, and large-scale compute clusters designed specifically to train robotic platforms.
AI infrastructure and compute +3.2Growing demand for compute is driving advanced packaging, liquid cooling deployments, and serving stack optimizations, while data center energy requirements face local public scrutiny.
AI-linked equities are broadly positive, with Advanced Micro Devices +4.69%, Super Micro Computer +4.54%, Palantir Technologies. -4.49% leading the tracked basket.
| Ticker | Company | Move |
|---|---|---|
| AMD | Advanced Micro Devices | +4.69% |
| SMCI | Super Micro Computer | +4.54% |
| PLTR | Palantir Technologies. | -4.49% |
| ARM | Arm Holdings plc | +3.92% |