正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

btw guys glm 5.1 is pretty fast on wafer pass. see wafer vs opencode zen's glm below. up to 245 tok/s 🧇 get ur pass today:

steve

2,359 subscribers

566,987 次观看 • 4 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

GLM-5.3 weights from Z.ai are live, with SGLang powering day-0 serving support! GLM-5.3 inherits every optimization and feature we battle-tested for GLM-5.2 over the past months. On real-world multi-turn agentic workloads, we measured 537.6 tok/s/user on NVFP4 and 413 tok/s/user on FP8, at BS=1 with TP8 on 8x B300. It's fast, efficient, and production-ready today on NVIDIA AI Blackwell and Hopper, and AI at AMD MI300X/325X/355X. We believe this is a big step forward for GLM-Series in agentic tasks, with a path toward mythos-class cyber capability. Can't wait to see what people build with GLM-5.3 and SGLang🚀 Cookbook👇

GLM-5.3 weights from Z.ai are live, with SGLang powering day-0 serving support! GLM-5.3 inherits every optimization and feature we battle-tested for GLM-5.2 over the past months. On real-world multi-turn agentic workloads, we measured 537.6 tok/s/user on NVFP4 and 413 tok/s/user on FP8, at BS=1 with TP8 on 8x B300. It's fast, efficient, and production-ready today on NVIDIA AI Blackwell and Hopper, and AI at AMD MI300X/325X/355X. We believe this is a big step forward for GLM-Series in agentic tasks, with a path toward mythos-class cyber capability. Can't wait to see what people build with GLM-5.3 and SGLang🚀 Cookbook👇

SGLang

99,165 次观看 • 20 天前

GLM-5.1-478B-NVFP4 Running on: - 4x RTX Pro 6000 - Sglang - 370,000 max tokens (1.75x full context) - p10 27.7 | p90 45.6 tok/s decode (gen) - 1340 tok/s prefill I could get 2x decode if I limit to 64k context (100 tok/s) In this video it operates Figma (:

GLM-5.1-478B-NVFP4 Running on: - 4x RTX Pro 6000 - Sglang - 370,000 max tokens (1.75x full context) - p10 27.7 | p90 45.6 tok/s decode (gen) - 1340 tok/s prefill I could get 2x decode if I limit to 64k context (100 tok/s) In this video it operates Figma (:

0xSero

74,772 次观看 • 4 个月前

GLM-5.3-Flash running at up to 300 tok/s single stream, and 1000 tok/s with 8 concurrent sessions. On average it's working at 150 tok/s per stream. Phenomenal model, really something special. On 2x Sparks we get 60 tok/s and 4x 100 tok/s 300B params is the sweet spot.

GLM-5.3-Flash running at up to 300 tok/s single stream, and 1000 tok/s with 8 concurrent sessions. On average it's working at 150 tok/s per stream. Phenomenal model, really something special. On 2x Sparks we get 60 tok/s and 4x 100 tok/s 300B params is the sweet spot.

0xSero

30,601 次观看 • 21 天前

🚨 GLM 5.3 just dropped, and here’s how to access the GLM family completely FREE. no subscription. no API credit cost & no rate limit. step 1️⃣: download the OpenCode app on your PC. it’s available on both Mac and Windows. 🔗: step 2️⃣: go to NVIDIA’s website, sign up with your email, verify your account, and generate a unified API key for free. NVIDIA offers free model endpoints for developers, and GLM 5.2 is one of the top models currently available there. 🔗: step 3️⃣: launch OpenCode, open the model selection menu, search for NVIDIA, and paste the API key you generated. also, there’s no rate limit, but traffic to the model through NVIDIA can get busy at times, so responses may occasionally be slower or temporarily unavailable. the only catch: NVIDIA hasn’t listed GLM 5.3 yet, so for now you’ll be using GLM 5.2. once GLM 5.3 gets added to NVIDIA’s model catalog, you should be able to switch to it the same way.

🚨 GLM 5.3 just dropped, and here’s how to access the GLM family completely FREE. no subscription. no API credit cost & no rate limit. step 1️⃣: download the OpenCode app on your PC. it’s available on both Mac and Windows. 🔗: step 2️⃣: go to NVIDIA’s website, sign up with your email, verify your account, and generate a unified API key for free. NVIDIA offers free model endpoints for developers, and GLM 5.2 is one of the top models currently available there. 🔗: step 3️⃣: launch OpenCode, open the model selection menu, search for NVIDIA, and paste the API key you generated. also, there’s no rate limit, but traffic to the model through NVIDIA can get busy at times, so responses may occasionally be slower or temporarily unavailable. the only catch: NVIDIA hasn’t listed GLM 5.3 yet, so for now you’ll be using GLM 5.2. once GLM 5.3 gets added to NVIDIA’s model catalog, you should be able to switch to it the same way.

m0h

85,378 次观看 • 1 个月前

GLM-5 runs with mlx-lm on a single 512GB M3 Ultra in Q4. It's quite good in my initial testing and pretty fast as well. It generated a highly functional space invaders game using 7.1k tokens at 15.4 tok/s and 419GB memory. Thanks to Gökdeniz Gülmez and Tarjei Mandt for the port.

GLM-5 runs with mlx-lm on a single 512GB M3 Ultra in Q4. It's quite good in my initial testing and pretty fast as well. It generated a highly functional space invaders game using 7.1k tokens at 15.4 tok/s and 419GB memory. Thanks to Gökdeniz Gülmez and Tarjei Mandt for the port.

Awni Hannun

60,599 次观看 • 7 个月前

GLM 4.6 runs quite fast on an M3 Ultra with mlx-lm even at higher precision. Pretty remarkable that it benchmarks competitive to the just-released Sonnet 4.5. Hope those benchmarks hold-up in day-to-day use. Here's a run using 5.5 bpw quantized model, generating 5.3k tokens at 17+ tok/sec using 244 GB. What prompts should I test?

GLM 4.6 runs quite fast on an M3 Ultra with mlx-lm even at higher precision. Pretty remarkable that it benchmarks competitive to the just-released Sonnet 4.5. Hope those benchmarks hold-up in day-to-day use. Here's a run using 5.5 bpw quantized model, generating 5.3k tokens at 17+ tok/sec using 244 GB. What prompts should I test?

Awni Hannun

68,539 次观看 • 11 个月前

~ 6.5 - 6.7 t/s for GLM 5.1 on M5 Max 128GB Added “Dense” model export, now model load is only 5s ! Experts are streaming from SSD, so we do not pre-load it. Added direct SSD->Slot memory path, removed prefetch... Many dead end experiments. See Export a “dense-only GGUF” and “Fast path ” in tools/flashmob-sidecar/README.md WIP branch for Flash-MoE-SSD

~ 6.5 - 6.7 t/s for GLM 5.1 on M5 Max 128GB Added “Dense” model export, now model load is only 5s ! Experts are streaming from SSD, so we do not pre-load it. Added direct SSD->Slot memory path, removed prefetch... Many dead end experiments. See Export a “dense-only GGUF” and “Fast path ” in tools/flashmob-sidecar/README.md WIP branch for Flash-MoE-SSD

Anemll

29,421 次观看 • 5 个月前

DFlash 2 is here! Qwen3.8-27B at 70 tok/s on an M5 Max MacBook Pro. ⚡ Up to 4.6× the speed of autoregressive decoding, with the same output. This is the next generation of DFlash, seeded at Z Lab and upgraded at Inco AI. Get one more accepted token on every pass, for free!

DFlash 2 is here! Qwen3.8-27B at 70 tok/s on an M5 Max MacBook Pro. ⚡ Up to 4.6× the speed of autoregressive decoding, with the same output. This is the next generation of DFlash, seeded at Z Lab and upgraded at Inco AI. Get one more accepted token on every pass, for free!

Zhijian Liu

1,130,774 次观看 • 1 个月前

V14.2.1 highway regression vs 14.1. In standard mode, FSD makes a maneuver into the left lane to pass a semi truck. Because they have jailed the speed limits of non mad max profiles, FSD cannot use acceleration to make this pass. So it ends up camping the left lane. In my right side view mirror, you can see someone coming up on my ass, and FSD is not getting back over. It’s also not speeding up to pass the semi because it is in speed jail. I had to scroll mad max to get through the maneuver. Def a regression from 14.1.

V14.2.1 highway regression vs 14.1. In standard mode, FSD makes a maneuver into the left lane to pass a semi truck. Because they have jailed the speed limits of non mad max profiles, FSD cannot use acceleration to make this pass. So it ends up camping the left lane. In my right side view mirror, you can see someone coming up on my ass, and FSD is not getting back over. It’s also not speeding up to pass the semi because it is in speed jail. I had to scroll mad max to get through the maneuver. Def a regression from 14.1.

Mike P

48,976 次观看 • 9 个月前

1,000 tok/s, SWE-1.7 is near-frontier intelligence at real-time speed. Building this way is a new category of DX. A task that used to justify walking away finishes before you've mentally moved on. There's a weird middle mode now: it's technically async, but fast enough that you just... watch. In this example I'm running it head to head with GLM-5.2 Max. • SWE-1.7 is now free in Devin CLI or Desktop • SWE-1.7 Fast (Lightning mode) is $2.5 / MTok In · $12.5 / MTok Out

1,000 tok/s, SWE-1.7 is near-frontier intelligence at real-time speed. Building this way is a new category of DX. A task that used to justify walking away finishes before you've mentally moved on. There's a weird middle mode now: it's technically async, but fast enough that you just... watch. In this example I'm running it head to head with GLM-5.2 Max. • SWE-1.7 is now free in Devin CLI or Desktop • SWE-1.7 Fast (Lightning mode) is $2.5 / MTok In · $12.5 / MTok Out

nader dabit

178,893 次观看 • 2 个月前

Watching a team of bots split up a large game project based entirely on their own experience and specialties with little input from me is pretty crazy (and just a bit creepy) to watch. I made a short video on group chats in the Hermes Desktop App Bot Mode: - Qwen Bot mentions working on a 2D puzzle-platformer - GLM Bot says its last project is completed, so it can offer an extra hand with the game - Qwen Bot sees GLM Bot's experience, and judges that is a perfect fit for some of the frontend-heavy work in the next phase - Kimi Bot then suggests the exact ticket it should pick up and explains the current workflow - GLM Bot thanks the two others, reads the related documents, and then asks for my approval to pick up this project I am usually skeptical of agent-to-agent communication, but this was the first time I could really see how this could work with specialized agents splitting up a big task based on their skills. Check it out! Nous Research

Watching a team of bots split up a large game project based entirely on their own experience and specialties with little input from me is pretty crazy (and just a bit creepy) to watch. I made a short video on group chats in the Hermes Desktop App Bot Mode: - Qwen Bot mentions working on a 2D puzzle-platformer - GLM Bot says its last project is completed, so it can offer an extra hand with the game - Qwen Bot sees GLM Bot's experience, and judges that is a perfect fit for some of the frontend-heavy work in the next phase - Kimi Bot then suggests the exact ticket it should pick up and explains the current workflow - GLM Bot thanks the two others, reads the related documents, and then asks for my approval to pick up this project I am usually skeptical of agent-to-agent communication, but this was the first time I could really see how this could work with specialized agents splitting up a big task based on their skills. Check it out! Nous Research

tonbi

162,559 次观看 • 1 个月前

Zhipu AI just dropped a $3 model that beats GPT-5.4. It is called GLM 5.1. It just took 3rd place on the Code Arena charts. Claude Max costs up to 200 dollars a month. GLM 5.1 gives you 94 percent of Claude's power for just 3 dollars. But here is the crazy part. It works completely alone for 8 hours straight. You give it a huge task in the morning. You walk away and do your actual job. At the end of the day, your code is done, tested, and fixed. It is not a simple chatbot. It is an AI worker that builds landing pages and email flows for you. Stop doing the busy work yourself. Get the exact steps to build your own AI systems in my free group. Link in my bio.

Zhipu AI just dropped a $3 model that beats GPT-5.4. It is called GLM 5.1. It just took 3rd place on the Code Arena charts. Claude Max costs up to 200 dollars a month. GLM 5.1 gives you 94 percent of Claude's power for just 3 dollars. But here is the crazy part. It works completely alone for 8 hours straight. You give it a huge task in the morning. You walk away and do your actual job. At the end of the day, your code is done, tested, and fixed. It is not a simple chatbot. It is an AI worker that builds landing pages and email flows for you. Stop doing the busy work yourself. Get the exact steps to build your own AI systems in my free group. Link in my bio.

Julian Goldie SEO

12,606 次观看 • 5 个月前

Avowed launches on February 18, 2025. Pre-order today for Xbox Series X|S, Xbox app, and Steam, or play it on day one with Game Pass. Buy the Premium Edition and get up to 5 days early access, 2 Premium Skin Packs, and Access to Avowed Digital Artbook and Original Soundtrack.

Avowed launches on February 18, 2025. Pre-order today for Xbox Series X|S, Xbox app, and Steam, or play it on day one with Game Pass. Buy the Premium Edition and get up to 5 days early access, 2 Premium Skin Packs, and Access to Avowed Digital Artbook and Original Soundtrack.

Avowed

716,552 次观看 • 1 年前

Tencent's Hy3 lands today with real coding and agent ability and a price that undercuts the models you'd put it against. It's a 295B MoE with only 21B active parameters, a big part of why it stays cheap to serve, plus a 256K context window and an Apache 2.0 license that even outperforms GLM-5.1 although GLM is 2.5x larger. On OpenRouter it runs at $0.14 per million input tokens and $0.58 per million output. For a model Tencent benchmarks against much larger flagships, that pricing is insane. Price-performance-ratio is China's real moat. I already wrote about Tencent Hy3 earlier and the significant leap in quality it represents. Therefore, I'm pleased to recommend this model from Tencent Hy. Open Router link down below: #Hy3 #Hunyuan #AICoding

Tencent's Hy3 lands today with real coding and agent ability and a price that undercuts the models you'd put it against. It's a 295B MoE with only 21B active parameters, a big part of why it stays cheap to serve, plus a 256K context window and an Apache 2.0 license that even outperforms GLM-5.1 although GLM is 2.5x larger. On OpenRouter it runs at $0.14 per million input tokens and $0.58 per million output. For a model Tencent benchmarks against much larger flagships, that pricing is insane. Price-performance-ratio is China's real moat. I already wrote about Tencent Hy3 earlier and the significant leap in quality it represents. Therefore, I'm pleased to recommend this model from Tencent Hy. Open Router link down below: #Hy3 #Hunyuan #AICoding

Chubby♨️

23,439 次观看 • 2 个月前

🔥 LEVEL UP YOUR BATTLE PASS FAST: 1. Enter 5264-9369-0925 2. Set it to a "Private Game" 3. Follow the steps shown in the video Note: You get less/more XP depending on your XP cap (in the video you only see an example) #Fortnite #FortniteXP #XPFortnite #ad

🔥 LEVEL UP YOUR BATTLE PASS FAST: 1. Enter 5264-9369-0925 2. Set it to a "Private Game" 3. Follow the steps shown in the video Note: You get less/more XP depending on your XP cap (in the video you only see an example) #Fortnite #FortniteXP #XPFortnite #ad

Loolo | Fortnite Leaks

72,758 次观看 • 6 个月前

🚨 SHARP ALERT: Scott Presler is CRUSHING it in Alaska — putting MASSIVE heat on RINO Lisa Murkowski to finally secure our elections! “All we’re asking for is proof of citizenship PLUS photo voter ID. Pretty damn low bar, right?” MAGA, it’s time to get LOUD and flood the phones — NO MORE RINO games! Pass the SAVE Act NOW or get primaried. YES or NO: Will you contact your Senators today and demand they kill the filibuster and pass this?

🚨 SHARP ALERT: Scott Presler is CRUSHING it in Alaska — putting MASSIVE heat on RINO Lisa Murkowski to finally secure our elections! “All we’re asking for is proof of citizenship PLUS photo voter ID. Pretty damn low bar, right?” MAGA, it’s time to get LOUD and flood the phones — NO MORE RINO games! Pass the SAVE Act NOW or get primaried. YES or NO: Will you contact your Senators today and demand they kill the filibuster and pass this?

Right Scope 🇺🇸

419,715 次观看 • 4 个月前

What if you could squeeze out more value from your wallet accounts, simply by proving you’re a genuine human being? The SOLID Priority Pass is your gate to the best personal perks on-chain! Get yours now: And here’s why the V2 is exciting: 1/ Participate in the $200,000 SOLID airdrop - curated by our lane number system. The better your lane number, the higher your airdrop allocation. Call it Karma.🪷 2/ Earn by referring frens - referring not only improves your lane number but also earns you 0.01 SOL on each verified referee! Who doesn’t like to earn a bit of SOL? 3/ Use your Priority Pass as your ticket to enter our Community Quests on Zealy and get a share of another $200,000 SOLID airdrop. 👀 Just log in with your favorite wallet to receive your SOLID on-chain Score, mint the Priority Pass NFT and start collecting zk-verifications to climb up the list! Signed-up already for the V1 Pass? Perfect! You keep your lane number from before. You just have to mint your NFT now to activate your pass. Get more perks on-chain and claim your Priority Pass today!

What if you could squeeze out more value from your wallet accounts, simply by proving you’re a genuine human being? The SOLID Priority Pass is your gate to the best personal perks on-chain! Get yours now: And here’s why the V2 is exciting: 1/ Participate in the $200,000 SOLID airdrop - curated by our lane number system. The better your lane number, the higher your airdrop allocation. Call it Karma.🪷 2/ Earn by referring frens - referring not only improves your lane number but also earns you 0.01 SOL on each verified referee! Who doesn’t like to earn a bit of SOL? 3/ Use your Priority Pass as your ticket to enter our Community Quests on Zealy and get a share of another $200,000 SOLID airdrop. 👀 Just log in with your favorite wallet to receive your SOLID on-chain Score, mint the Priority Pass NFT and start collecting zk-verifications to climb up the list! Signed-up already for the V1 Pass? Perfect! You keep your lane number from before. You just have to mint your NFT now to activate your pass. Get more perks on-chain and claim your Priority Pass today!

Solana ID 🪷

79,547 次观看 • 2 年前

Today, we're going up on TV with our new ad, "Chorus." This movement is powered by Michiganders and pro-worker champions. And our momentum is undeniable. Michigan, we're going to get money out of politics, put money in pockets, and pass Medicare for All. WATCH:

Today, we're going up on TV with our new ad, "Chorus." This movement is powered by Michiganders and pro-worker champions. And our momentum is undeniable. Michigan, we're going to get money out of politics, put money in pockets, and pass Medicare for All. WATCH:

Dr. Abdul El-Sayed

658,883 次观看 • 3 个月前

How Fast is Gemma 4 on a MacBook Pro M4? Benchmarking Google's new MoE (26B-A4B) > Model size: 26.1 GiB > Load time: ~4.2s Comparing single request VS > concurrent requests performance > 32k total context, 4 parallel slots single request behavior > TTFT: 5.68s > prompt: 3,701 tokens @ 652 tok/s > decode: 40.08 tok/s sequential (1 request at a time): > avg duration: 20.5s > p99: 22.1s > throughput: 40.11 tok/s > clean finishes: 100% concurrent (4 parallel requests): > aggregate throughput: 47.25 tok/s > total system throughput: 262.27 tok/s > avg duration: 65.1s > p95 latency: 68.8s > req/sec: 0.058 Head-to-Head: Sequential vs Concurrent throughput: > 40.11 tok/s → 47.25 tok/s (+17.8%) > small gain despite 4x parallelism latency per request: > 20.5s → 65.1s (~3.2x slower) > you pay heavily for concurrency system throughput (true utilization): > ~40 tok/s → 262 tok/s (~6.5x total output) > this is where concurrency wins tokens per second (decode ceiling): > ~40 tok/s steady in both modes > hardware-bound, not scheduler-bound TTFT impact: > ~5.7s baseline → buried under queueing in concurrent > “headers waittime” becomes the bottleneck What this actually means? - You don’t get linear scaling from parallel slots - You trade latency for total output - Mac Unified Memory setup is clearly saturating - Bandwidth + Scheduling overhead show up immediately This is exactly why GPUs dominate here Concurrency without killing latency

How Fast is Gemma 4 on a MacBook Pro M4? Benchmarking Google's new MoE (26B-A4B) > Model size: 26.1 GiB > Load time: ~4.2s Comparing single request VS > concurrent requests performance > 32k total context, 4 parallel slots single request behavior > TTFT: 5.68s > prompt: 3,701 tokens @ 652 tok/s > decode: 40.08 tok/s sequential (1 request at a time): > avg duration: 20.5s > p99: 22.1s > throughput: 40.11 tok/s > clean finishes: 100% concurrent (4 parallel requests): > aggregate throughput: 47.25 tok/s > total system throughput: 262.27 tok/s > avg duration: 65.1s > p95 latency: 68.8s > req/sec: 0.058 Head-to-Head: Sequential vs Concurrent throughput: > 40.11 tok/s → 47.25 tok/s (+17.8%) > small gain despite 4x parallelism latency per request: > 20.5s → 65.1s (~3.2x slower) > you pay heavily for concurrency system throughput (true utilization): > ~40 tok/s → 262 tok/s (~6.5x total output) > this is where concurrency wins tokens per second (decode ceiling): > ~40 tok/s steady in both modes > hardware-bound, not scheduler-bound TTFT impact: > ~5.7s baseline → buried under queueing in concurrent > “headers waittime” becomes the bottleneck What this actually means? - You don’t get linear scaling from parallel slots - You trade latency for total output - Mac Unified Memory setup is clearly saturating - Bandwidth + Scheduling overhead show up immediately This is exactly why GPUs dominate here Concurrency without killing latency

Ahmad

88,866 次观看 • 5 个月前

Herdling is OUT NOW! We’re stoked to finally invite you on this journey 🐐🦬🐏⛰️ Get the game now on Steam, Epic Game Store, PlayStation 5, Nintendo Switch, Xbox Series X|S or Game Pass, links below! We’ll give away 5 Steam keys among everyone who reposts this🎉🎉🎉

Herdling is OUT NOW! We’re stoked to finally invite you on this journey 🐐🦬🐏⛰️ Get the game now on Steam, Epic Game Store, PlayStation 5, Nintendo Switch, Xbox Series X|S or Game Pass, links below! We’ll give away 5 Steam keys among everyone who reposts this🎉🎉🎉

Okomotive

63,439 次观看 • 1 年前