What the local brain observed in archived AI discourse: quota and limit resets with the times they were seen and the times they apply, model releases, dev tools, and terms rising above their baseline. Every row links its evidence.
@thsottiaux: There were reports of usage draining faster than anticipated for some users over the weekend. We've identified approximately 40,000 accounts that had an inconsistency on which usage limits where the actual usage that was applied did not correspond to their actual
observed
1 source
@thsottiaux — There were reports of usage draining faster than anticipated for some users over the weekend. We've identified approximately 40,000 accounts that had an inconsistency on which usage limits where the
Model releases
19 in window
Model releaseFlora
Flora open-weights robotics model suite released
Flora is introduced as an open suite of robotics models to parse physical environments, with weights open-sourced today.
observed applies today
1 source
@drjimfan — We built a robotics data engine with Astra-level performance at 125x lower cost. And we are open-sourcing the weights today! Introducing Flora, a suite of open models to parse the physical world and
Model releaseMistral
Mistral release announced
@artificialanlys: Mistral has released new Speech models! Voxtral Small Transcribe 3 takes the #3 spot on the Artificial Analysis Speech to Text leaderboard, and Voxtral 3 TTS enters the top 12 in both Text to Speech arenas @MistralAI has released Voxtral 3:
observed
1 source
@artificialanlys — Mistral has released new Speech models! Voxtral Small Transcribe 3 takes the #3 spot on the Artificial Analysis Speech to Text leaderboard, and Voxtral 3 TTS enters the top 12 in both Text to Speech
Model releaseMistral
Mistral introduces Voxtral 3 audio models
Mistral introduced Voxtral 3 transcribe and speech models powering its new voice agents API.
observed
1 source
@mistralai — Introducing Voxtral 3: state-of-the-art Transcribe and Speech models powering Mistral’s new voice agents API. 🧵
Model releasePocket TTS
Kyutai releases smaller Pocket TTS models down to 1.9M parameters
Kyutai Labs released smaller versions of Pocket TTS down to 1.9M parameters, including an 18.2M model matching the original quality, a 6.9M model, and a 1.9M model running 103x faster.
observed
1 source
@kyutai_labs — Pocket TTS has been getting updates those last few days. While it was ~100M parameters, we release smaller versions, down to 1.9M: an 18.2M model with the quality of the original, a 6.9M model, and a
Model releaseMercury 3
Inception releases Mercury 3 model
A user post notes Inception released Mercury 3, describing it as faster with comparable benchmark performance to Luna and Haiku.
observed
1 source
@deedydas — Inception just dropped Mercury 3, a model 4-8x faster than Luna / Haiku for the same price and benchmark performance. In real world use, OpenRouter shows that it’s effectively 2-4x faster and 2-4x
Model releaseMuse Glimmer
Technical report published on Muse Glimmer 30B open-weight model
A technical report was published detailing Muse Glimmer, a 30B open-weight model for agentic tasks on consumer hardware.
observed
1 source
@aiatmeta — Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware. It runs at practical speeds on a Mac or PC with a single consumer GPU at
Model releaseFLUX
FLUX release announced
@bfl_ai: Introducing FLUX 3 Max. Fastest in the west. Fast enough to try anything. And everything. Generate a 10-second video in 12 seconds. Tied for #1 in image-to-video. 22× faster than Seedance 2.5 at the same quality. FLUX 3 is a model designed
observed
1 source
@bfl_ai — Introducing FLUX 3 Max. Fastest in the west. Fast enough to try anything. And everything. Generate a 10-second video in 12 seconds. Tied for #1 in image-to-video. 22× faster than Seedance 2.5 at the
Model releaseDeepSeek V4.1 Flash 1017
DeepSeek releases DeepSeek V4.1 Flash 1017
DeepSeek released DeepSeek V4.1 Flash 1017, an updated open-weights model scoring 45 on the Artificial Analysis Index.
observed
1 source
@artificialanlys — DeepSeek V4.1 Flash 1017 scores 45 on the Artificial Analysis Intelligence Index, improving on DeepSeek V4.1 Flash by 6 points, and is the equal #2 open-weights model @deepseek_ai has released
Model releaseMiniMax
OpenRouter reveals stealth model Space Bunny as MiniMax M3.1 Flash
OpenRouter revealed that stealth model Space Bunny has returned as MiniMax M3.1 Flash.
observed applies today
1 source
@openrouter — Stealth model reveal 🐰 Space Bunny was OpenRouter’s #1 model for 10 straight days, processing 56.7T tokens! The most of any stealth model we’ve run. Space Bunny hopped off OpenRouter on Oct 5. It's
Model releaseOrcaCyber Nano 1.0
OrcaCyber Nano 1.0 released as open-weight model
@orcarouter announces the open-weight release of OrcaCyber Nano 1.0, a 27B parameter vision model with 262K context in the OrcaCyber series.
observed applies now
1 source
@orcarouter — 🚨 Emergency release: OrcaCyber Nano 1.0 is now open-weight. Part of our OrcaCyber model series: Nano, Zero, Ultra. Based on our early evals, Nano is the most capable local model in its under-16GB
Model releaseMiniMax
MiniMax release announced
@arena: MiniMax-M3.1-Flash by @MiniMax_AI is live in the Code Arena: WebDev. It debuts at #12 among open models with 1588 pts (#32 overall). It also lifts MiniMax to #12 among all labs! MiniMax-M3.1-Flash is a +106 pt increase from the previous var
observed applies from 11 minutes to 72 seconds
3 sources
@arena — MiniMax-M3.1-Flash by @MiniMax_AI is live in the Code Arena: WebDev. It debuts at #12 among open models with 1588 pts (#32 overall). It also lifts MiniMax to #12 among all labs! MiniMax-M3.1-Flash is
@aibattle_ — MiniMax is preparing to release a Pro model soon According to earlier reports from The Information and Reuters, the model is expected to have a total of 2.7T parameters
@0x0sojalsec — MiniMax H3 just got 9× faster On an RTX 4090. Locally video generation shrink down From 11 minutes to 72 seconds on 16 GB GPUs Open-source New ComfyUI node drops MiniMax H3 character video generation
@askvenice — DeepSeek V4.1 Flash just got a serious upgrade. @deepseek_ai’s 1017 outscores leading open-source and frontier models on nearly every benchmark. Same model in the Venice app, simply upgraded and
@bindureddy — 🚨 DeepSeek Just Dropped The Best And Cheapest Open Source Model In The World - scores on top of the open-source leaderboard - beats Muse Wonder, Haiku, and Luna - cheapest in terms of cost per task
@ollama — ✈ DeepSeek-V4.1-Flash-1017 is on Ollama! US-based hosting with zero data retention. It's strong on agentic tasks while being cheaper per task. Try it on Ollama Cloud: ollama run
@deepseek_ai — 🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our
Model releaseIndra
Indra 1M parameter OCR model launched
Indra is a 1M parameter OCR model launched this week, described as 15x faster than Tesseract on CPU with improved accuracy.
observed this week
1 source
@vikparuchuri — This week, we launched Indra, a 1M parameter OCR model that's 15x faster than Tesseract on CPU and much more accurate. I still can't believe how good this model is.
Model releaseBeam
Beam open-weights model released on Hugging Face
Reflection's Beam model, an open weights model with 501B total and 23B active parameters, is released on Hugging Face.
observed
1 source
@artificialanlys — Reflection’s first model, Beam, scores 30 on the Artificial Analysis Intelligence Index, placing it #2 among US open weights models Beam is an open weights model with 501B total and 23B active
Model releaseMuse Spark Open
Meta releases Muse Spark Open open-weights model
Meta released Muse Spark Open, an 816B (42B active) open-weights model scoring 39 on the Artificial Analysis Intelligence Index.
observed
1 source
@artificialanlys — Meta has released Muse Spark Open, an 816B (42B active) open-weights model that scores 39 on the Artificial Analysis Intelligence Index, making it the #1 US open-weights model Muse Spark Open (xhigh)
Model releaseClaude Opus
Claude Opus 5.5 fast mode enters general availability
Opus 5.5 fast mode is no longer in research preview on Claude Platform and Microsoft Foundry, removing the waitlist with up to 2.5x speed at $8 / $40 per million tokens.
observed
1 source
@claudedevs — Opus 5.5 fast mode in the Claude Platform is no longer in research preview, and we’ve removed the waitlist! It’s the same Opus 5.5 but up to 2.5x faster, at $8 / $40 per million tokens. It's also
Model releaseMuse Spark Open
Meta releases Muse Spark Open
Meta released Muse Spark Open, an 816B open model supporting text, images, video, and audio with 1M context, available via Unsloth GGUFs.
observed
2 sources
@unslothai — Meta releases Muse Spark Open, a new SOTA open model. Run on a Mac Studio via Dynamic 2-bit Unsloth GGUFs. Muse-Spark-Open-816B-A42B supports text, images, video and audio input with 1M context. GGUF
@finkd — Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple.
Model released1-ultra
Liquid AI releases d1-ultra decision model
Liquid AI has released d1-ultra, a multimodal decision model supporting 44 languages with evaluations on the Artificial Analysis Decision Index.
observed applies today
1 source
@liquidai — Today we release 𝚍𝟷-𝚞𝚕𝚝𝚛𝚊, our most powerful decision model. > built for hard decision problems. > d1-ultra is multimodal > supports 44 languages > #1 on the @ArtificialAnlys Decision Index
Model releaseNVIDIA
NVIDIA releases Nemotron-3.5-Bolt and Nemotron-3.5-Super-VL
A post from @unslothai details NVIDIA models Nemotron-3.5-Bolt-74B-A9B and Nemotron-3.5-Super-VL-120B-A12B, with GGUF weights available.
observed
1 source
@unslothai — NVIDIA releases 2 new models: Nemotron-3.5-Bolt and Nemotron-3.5-Super-VL. Nemotron-3.5-Bolt-74B-A9B runs on 24GB RAM. It delivers top-tier performance for its size. Super-VL-120B-A12B runs on 82GB
Dev tools
13 in window
Dev toolOllama
Ornith 1.5 122B available on Ollama
Ollama announced that Ornith 1.5 122B is available to run locally on devices using ollama run.
observed
1 source
@ollama — Ornith 1.5 122B is available on Ollama! Run it all locally on your device. ollama run ornith-1.5:122b
Dev toolDeepSeek
DeepInfer inference framework open sourced
DeepSeek open-sourced DeepInfer, an in-house inference framework powering DeepSeek-V4.1-Flash at scale.
observed
1 source
@deepseek_ai — 🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to
Dev toolOllama
Ollama adds MiniMax M3.1 Flash support
MiniMax M3.1 Flash is available on Ollama and Ollama Cloud with US hosting, zero data retention, and a 1M-token context window.
observed
1 source
@ollama — MiniMax M3.1 Flash is on Ollama! 🔒 Hosted in the US and with zero data retention. Fast coding and agentic workflows, with image understanding and a 1M-token context window. Try it on Ollama Cloud
Dev toolUnsloth
Unsloth releases dynamic GGUFs and llama.cpp implementation for DS4.1-Flash
Unsloth developed a llama.cpp implementation and dynamic GGUFs for DS4.1-Flash, enabling precision swaps across MXFP8, Q6_K, Q5_K, and IQ4_NL.
observed
1 source
@danielhanchen — We worked on a llama.cpp impl for DS4.1-Flash + made Unsloth dynamic GGUFs for it! The engram / PLE can be swapped out from MXFP8 to Q6_K to Q5_K and IQ4_NL precision - we found Q6_K approx = MXFP8
Dev toolvLLM
vLLM update announced
@vllm_project: 🚀 vLLM v0.32.0 is out! 814 commits · 363 contributors · 121 first-timers 🧭 New /v1/systemone structured-decision endpoint + generic hf tool/reasoning parser 🗜️ Low-precision KV everywhere: NVFP4 on SM8x/90/12x, FP8 KV default for GLM-5.x, 4
Search and Fetch APIs are now open to everyone, priced at $0.75 per 1,000 calls each endpoint with documentation available.
observed applies now open
1 source
@deepseek_ai — 🔍 Search & Fetch APIs are now open to everyone! Powered by our in-house search engine, serving hundreds of millions of requests daily worldwide. 💰 Each endpoint: $0.75 per 1,000 calls. 📖 Docs
Dev toolClaude Code
Claude Code adds gzip compression to conversation turns
Claude Code now gzips conversation turns to reach the model faster, alongside faster cloud session startup.
observed applies now
1 source
@claudedevs — We keep improving performance across Claude, and you may have noticed it starts responding sooner! Claude Code now gzips the conversation it sends on every turn, so it reaches the model faster. Cloud
Dev toolHermes Agent
Hermes Agent partners with Nvidia RTX Spark for Windows laptops
Hermes Agent is a launch partner for Nvidia RTX Spark laptops, integrating into the Windows agent setup flow for local and cloud workflows.
observed
1 source
@nousresearch — Hermes Agent is a launch partner for @nvidia's RTX Spark across the new generation of laptops. Pick it in the built-in @Windows agent setup flow, run the new generative onboarding, and get straight to
Dev toolOllama
Ollama adds NVIDIA Nemotron 3.5 Bolt 74B-A9B 24GB support
Ollama announces NVIDIA Nemotron 3.5 Bolt 74B-A9B 24GB is available on Ollama, using native 2-bit quantization to run locally in 24GB of memory for everyday agentic tasks.
observed
1 source
@ollama — NVIDIA Nemotron 3.5 Bolt 74B-A9B 24GB is now on Ollama This model vision-language model (VLM) is designed to run everyday agentic tasks locally, using a native 2-bit quantization to fit in 24GB of
Dev toolCursor
Cursor update announced
@cursor_ai: Muse Wonder from Meta Superintelligence Labs is now available in Cursor! It scores 46.1% on CursorBench at default effort, the highest of any Muse model. Priced at $4/M input and $20/M output tokens.
observed applies now
1 source
@cursor_ai — Muse Wonder from Meta Superintelligence Labs is now available in Cursor! It scores 46.1% on CursorBench at default effort, the highest of any Muse model. Priced at $4/M input and $20/M output tokens.
Dev toolClaude
Code Review generally available for Claude Team and Enterprise plans
Code Review is now generally available for Claude Team and Enterprise plans, offering faster reviews and configurable Quick, Standard, or Deep review levels per repo.
observed applies now
1 source
@claudedevs — Code Review is now generally available for Claude Team and Enterprise plans. Reviews are now faster, catch more of the bugs that matter, and admins can pick from Quick, Standard, or Deep review levels
Dev toolOpenRouter
OpenRouter update announced
@openrouter: Today, we are launching OpenRouter Armor in beta. Tool calls allow agents to interact with the world, and regex-based filtering often isn't enough to detect unwanted tool calls. Armor uses @typesafeai 's Jev model to guard against out of po
observed applies today
1 source
@openrouter — Today, we are launching OpenRouter Armor in beta. Tool calls allow agents to interact with the world, and regex-based filtering often isn't enough to detect unwanted tool calls. Armor uses @typesafeai
Dev toolHermes Agent
Hermes Agent update announced
@teknium: Hermes Agent v0.22.0 is now live! The highlights: A much smoother easier to use onboarding - you don't even need to hook up a provider to get started - just install and start working! Connectors - much easier to connect with any app, just c
observed applies now
1 source
@teknium — Hermes Agent v0.22.0 is now live! The highlights: A much smoother easier to use onboarding - you don't even need to hook up a provider to get started - just install and start working! Connectors
Anthropic commits shares to new Anthropic Public Benefit Fund
The company is committing 5% of its shares to create the Anthropic Public Benefit Fund, dedicated entirely to public benefit.
observed
1 source
@anthropicai — We’re committing a portion of Anthropic’s shares to the Anthropic Public Benefit Fund, a new wholly owned entity dedicated entirely to the public benefit. It will start off with 5% of our shares. For
PricingFLUX
FLUX 3 Video price reduced by 50 percent
FLUX 3 Video generation is up to 2x faster and costs 50% less for native 2K and 4K generation following compute optimizations.
observed applies now
1 source
@bfl_ai — FLUX 3 Video is now up to 2x faster and costs 50% less for generating in native 2K and 4K. We made optimizations so it uses less compute to serve. The new pricing and speed improvements reflect that.