Pinned Tweet
I just created a Ko-fi page! If youβd like to support my work (or just buy me a coffee), check it out here: ko-fi.com/testtm
Will help me and cats a lot :( and bring me one step closer to my dream of creating my very own large language model (LLM). Every bit helps β₯οΈ
Will help me and cats a lot :( and bring me one step closer to my dream of creating my very own large language model (LLM). Every bit helps β₯οΈ
5
3
30
27,851
My bet: LongCat-2.5 was trained on Huawei Ascend 910C.
The chip is never named, but:
β’ "AIV-based comms" to avoid "AI-CPU task expansion" = HCCL lingo
β’ <80GB per device = 64GB per 910C die
β’ FSDP/EP sized to the supernode = CloudMatrix-style SuperPoD
The chip is never named, but:
β’ "AIV-based comms" to avoid "AI-CPU task expansion" = HCCL lingo
β’ <80GB per device = 64GB per 910C die
β’ FSDP/EP sized to the supernode = CloudMatrix-style SuperPoD
Introducing LongCat-2.5, our latest natively multimodal agentic model with 1.6T total parameters and ~48B active.
Built for long-horizon agentic tasks, it brings stronger visual understanding and agentic coding to work ranging from complex coding projects to visual coding such 3D tasks.
Learn more about LongCat-2.5: longcat.ai/blog/longcat-2β¦
2
0
10
923
Heads-up: could also be 910B on Atlas A2, since some of the release mentions A2. But the whole parallelism design is built around a big supernode domain, which only really pays off on A3/CloudMatrix. Still leaning 910C.
1
0
3
81
That cat is LOONG! and it is also EPIC release!
Introducing LongCat-2.5, our latest natively multimodal agentic model with 1.6T total parameters and ~48B active.
Built for long-horizon agentic tasks, it brings stronger visual understanding and agentic coding to work ranging from complex coding projects to visual coding such 3D tasks.
Learn more about LongCat-2.5: longcat.ai/blog/longcat-2β¦
0
0
11
343
Honestly very cool that more competitors show up. hope one will make good cheap accelerators for local ai :)
ig 2030 the earliers, cus Memory is the biggest problem. but hope CXMT gona scale fast.
ig 2030 the earliers, cus Memory is the biggest problem. but hope CXMT gona scale fast.
China's Iluvatar CoreX claims its Tiangai 300 GPU beats NVIDIA's H200 in its own live demo. π wccf.tech/1lgsm
0
0
1
159
Google TPUs like space!
Our TPUs just phoned home... from space.
In early October, we sent four TPUs into orbit to test whether our AI hardware could operate in space. The initial goal was to see if our chips could turn on, process a prompt, and transmit a response back to Earth.
This week, we got our answer.
0
0
4
118
And ofc nothing for Plus....
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and wonβt be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. Weβre also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
0
0
0
78
so cool, fat cat!
Today, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity.
π» Built to build: Le Chonk ranks #1 on @cognitionβs FrontierCode among open models, and sets a new benchmark for Pareto efficiency on Built to defend: State-of-the-art in cyberdefense, put through the paces by the world's leading cyber vendors, including @CrowdStrike, @cloudflare, @rubrik, @zscaler, and others.
Built for sovereignty: Make it yours. Download the weights on @huggingface , customize it with Mistral Forge, and deploy it wherever you need it.
Read more from our team and customers π:
0
0
5
255
SO COOL! Gemini 4 Carbon released!
Introducing Gemini 4 Carbon, it's a big jump over Argon which we released just 2 weeks ago.
Gemini 4 Carbon is SOTA on cyber, long horizon tasks, ranks #1 on Artificial Analysis V5, significantly improves on usability in domains like coding, and is more aligned.
0
0
3
131
testtm retweeted
@test_tm7873 The purpose of Artificial Analysis is to keep the Chinese out, Antrhropic in, and OpenAI down.
0
1
2
293
For me, Artificial Analysis is no more trust worthy. This update was made to just make openai higher, Chinese models lower.
And it's not even thier first time, updates like that happend many times. All with the same result...
There is no way 6.1 Terra is better than Fable 5.1, or trustworthy benchmark leader board look at Vals Ai leader board.
And it's not even thier first time, updates like that happend many times. All with the same result...
There is no way 6.1 Terra is better than Fable 5.1, or trustworthy benchmark leader board look at Vals Ai leader board.
Announcing Artificial Analysis Intelligence Index v5, featuring 2 new benchmarks and 3 upgrades. For the first time, the majority of the Index is built on private datasets
The Artificial Analysis Intelligence Index is our composite measure of model capabilities across tasks sage today. v5 is our biggest upgrade since v4 launched in January. With v5, 60% of the Index is now based on evaluations with private test sets or held-out answers.
Alongside Intelligence Index v5, weβre launching the Artificial Analysis Progress Index v1, which tracks model capability over time on one continuous scale.
New to the Intelligence Index:
β€ FrontierCode 2.0: Cognition's newly-launched private benchmark of real-world coding tasks, graded on whether a repository maintainer would merge the agent's change
β€ Terminal-Bench-Science: 70 agentic scientific research tasks, run in a terminal environment
Upgraded:
β€ HLE-Diamond (replacing Humanity's Last Exam): a refined 1,000-question version of Humanity's Last Exam, split evenly between reasoning and knowledge questions
β€ AA-Omniscience v1.1: An updated dataset with saturated and public set questions removed, new questions added, and ambiguous or erroneous questions revised
β€ CritPt v2: a private, 40-question hard subset of CritPt v2, revised by the CritPt authors and run with a code-execution tool
Staying in the Index: AA-Briefcase, AutomationBench-AA, GDP.pdf and Terminal-Bench 4.0
Removed from the Index: GDPval-AA v2.1, SciCode, and AA-LCR v1.1
Key results:
β€ Claude Opus 5.5 (max) leads at 58, ahead of GPT-6 Astra (max, 57) and GPT-6.1 Sol (max, 55): Anthropic and OpenAI hold the top five models. Gemini 4 Argon (high) is next at 50
β€ The open weights gap widens from 3 points to 12: Kimi K3.1 Pro (weights not yet released) is the top open weights model at 46. Among released weights, DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) lead at 32 - 26 points behind the leader and below Claude Haiku 5.5 (max, 36)
β€ Cost per task varies by more than 13x among the leading models: of the 10 highest-scoring models shown, GPT-6.1 Terra (max) is the cheapest at $0.96 per Intelligence Index task and Claude Fable 5.1 (max) the most expensive at $13.04
23
5
274
23,101
Massive release honestly! so cool!
Introducing π»π΅πΌπΉ: our new generation of Liquid Foundation Models. With a new tri-hybrid architecture and new multimodal encoders, the LFM3 family is the ideal model suite for specialized intelligence.
> 6 model instances, across text, audio, vision, and omni modalities, gendels.
> π»π΅πΌπΉ-π·.π½π±: text, 256k context window.
> π»π΅πΌπΉ-π π»-πΈπ±: vision-language model with video-input support for spatial understanding
> π»π΅πΌπΉ-π°ππππ-πΈπ±: tool calls, voice chat, reasoning in 18 languages.
> π»π΅πΌπΉ-πΎπππ-πΈπ±: an omni model combines vision, audio, and language.
> ππ·-πΈπ± and ππ·-πΎπππ-πΈπ±: decision models with typed outputs and calibrated probabilities in a single pass. Omni adds audio input.
> fine-tune on 1 GPU.
> a new tri-hybrid architecture
> architecture optimized for 14 processor classes across GPUs, CPUs, and NPUs.
> π»π΅πΌπΉ-π π»-πΈπ± is a beast. It performs competitively with Gemma 4 12B (6x larger), and Gemini 3.5 Flash Lite, and GPT-5.6 Luna, with 1.9B parameters.
> π»π΅πΌπΉ-π·.π½π± is #1 on the Pareto frontier of device intelligence πππππππ.ππππππ.ππ
Every modality for specialized intelligence. Pick the one that fits your use case, customize it, and run it on-device.
1/ π§΅
0
0
16
421
PEAK RELEASE FROM GDM!!
Introducing Genie 4 from @GoogleDeepMind, our most realistic and interactive world model yet π
Now you can use natural language to dynamically steer the world and its soundscape in real-time, whether you're changing the weather, moving the camera, or introducing new elements to ow does the model do that?
Behind the scenes, Gemini acts as the "brain" of the simulation. It translates your requests into actions, plans out complex events, watches the scene unfold to keep track of what's happening, and adjusts as you change the story. Genie 4 also introduces ambient audio and real-time interactive human dialogue that adapts on the fly.
2
0
9
421
Ok officialy Claude is no longer best for coding!
Kimi 3.1 Pro is now SOTA on Terminal Bench 4.0!
Kimi 3.1 Pro is now SOTA on Terminal Bench 4.0!
2
1
26
1,646
