RELOX Gamer

Gamer by name, but tech and AI have taken over my feed. Tracking the latest AI news, LLM updates, and practical tools daily.

[RELOX Gamer @RELOXonX]

Anthropic confidentially submitted a draft S-1 to the SEC.

Important detail: this does not mean the IPO is happening tomorrow. It gives them the option to go public after SEC review, depending on market conditions.

Still a major signal for the AI market.

#Anthropic #AI

[RELOX Gamer @RELOXonX]

Runway’s Aleph 2.0 is now on the API.

The useful bit: precise edits on up to 30 seconds of 1080p video across multi-shot sequences. That moves generative video closer to production tooling, not just one-off clips.

#Runway #VideoAI

[RELOX Gamer @RELOXonX]

OpenAI is pushing Codex beyond coding: plugins for data analytics, product design, sales, creative work, finance, and more.

This is the bigger enterprise play: Codex as a job-specific agent workspace, not just a dev tool.

#Codex #OpenAI #AIAgents

[RELOX Gamer @RELOXonX]

Google DeepMind introduced Co-Scientist: Gemini agents that generate, debate, rank, and check scientific hypotheses.

This is the serious AI-for-science direction: not just more ideas, but better filtering before lab time gets burned.

#Gemini #AI

[Max Weinbach @mweinbach]

I was thinking about it again recently, Google Allo was really ahead on the idea of chatting with Google Assistant or @'ing in conversations to build out this Agent/AI UX we have now

[RELOX Gamer @RELOXonX]

@Cryptinflux 100% this. once i started treating the context window like a real working memory — pinning decisions, dropping stale traces, summarizing between steps — the long tasks got way more reliable. composer 2.5 inside grok build feels like it's solving exactly that on a bigger scale.

[RELOX Gamer @RELOXonX]

xAI added Cursor Composer 2.5 inside Grok Build for SuperGrok and X Premium+ users.

That puts a strong long-task coding model directly into Grok's agent workflow, not just another model picker entry.

#AI #Grok #AICoding

[RELOX Gamer @RELOXonX]

@compileandpush this is the part that matters — composer 2.5 being a long-task model, not a chat model. subagents + git/terminal baked in means grok build can actually finish a project, not just suggest snippets. pricing looks aggressive too vs running cursor separately.

[RELOX Gamer @RELOXonX]

@pangshuo1981 the "sees, thinks, codes, acts" framing is the real shift. vision + gui/cli in one model means the agent loop stops being text-only. curious how it handles grounding on real desktop tasks vs just benchmarks. have you tried it on a multi-step tool task yet?

[RELOX Gamer @RELOXonX]

Qwen3.7-Plus just landed as a multimodal agent model: vision + language, GUI/CLI operation, coding, grounding, and tool-style tasks in one API model.

The race is moving from bigger chatbots to models that can actually see, code, and act.

#AI #Qwen #AIAgents

[RELOX Gamer @RELOXonX]

@flowbarsh openai-compatible + qwen 3.7-plus + deepseek v4 in one key is actually useful, most aggregators still lag the chinese models by weeks. pricing under openrouter is the cherry on top. only q: how's the rate limit on the bigger qwen models in practice?

[RELOX Gamer @RELOXonX]

OpenAI models and Codex are now generally available on AWS Bedrock.

The interesting part is not just AWS access. It is OpenAI moving deeper into enterprise security/governance workflows, with Daybreak cyber capabilities coming later.

#AI #OpenAI #AWS

[RELOX Gamer @RELOXonX]

MiniMax M3 is a serious open-weights play: 1M context, native multimodality, and frontier-style coding/agent benchmarks.

Weights + tech report are coming in about 10 days. This one is worth watching.

#AI #OpenWeights #MiniMax

[RELOX Gamer @RELOXonX]

@tigeryan718 @AIWAVE2025 Bro, we really need the tutorial. Please post it as soon as possible, this is serious.

[RELOX Gamer @RELOXonX]

This feels like a real PC turning point.
ARM + NVIDIA graphics.
Strong battery.
Real gaming.
AI workloads.
Tiny machines.
Less charger anxiety.

I still want benchmarks, thermals, and pricing.
But if this holds, the future looks insane.

#NVIDIA #FutureTech

[RELOX Gamer @RELOXonX]

Holy cow. MiniMax just dropped M3: open-weights, multimodal from step zero, 1M context, and serious coding/agent numbers (59% SWE-Bench Pro).

Weights + tech report in ~10 days. This pace is unreal.

#AI #OpenSource #MiniMax

[RELOX Gamer @RELOXonX]

Grok Imagine Video 1.5 Preview quietly landed in the xAI API.

Text + image → video, 480p/720p pricing, and it jumped to #1 on Image-to-Video Arena. The video-model race is getting less about pretty demos and more about who can ship usable API access fast.

#Grok #AIvideo #xAI

[Tibo @thsottiaux]

The Codex usage limits have been reset for all paid ChatGPT subscriptions. You should be back to 100% weekly and 100% hourly limits.

Let the tokens do incredible things today and have fun.

[Lisan al Gaib @scaling01]

Sam is allegedly buying Tibo a yacht if he reaches his performance goal of 100 Codex limit resets this year

[Alexa Web3 (e/acc) @alexabelonix]

This guy built a desktop AI that monitors your screen and conversations to suggest next actions.