Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Qwen3.8-Flash-Next from Qwen has day-0 support in Atomic Agent! 125B main model, 51B N-gram embeddings, 6B activated per token and 62.5 on SWE-bench Pro. We gave it a folder and it checked every file to sort them by content. Total: 25 steps + 230K tokens ≈ $0.07

atomicagent_io's profile picture

Atomic Agent

2,379 subscribers

43,968 görüntüleme • 3 gün önce •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

Qwen3.8-Max became the brain of Atomic Agent, Hermes and OpenClaw. We gave the same task: Turn a photo of a hand-drawn floor plan into an interactive 3D walkthrough of that apartment and open it in the browser. Outputs: – Atomic Agent: 66 min, 557K tokens, $2.01 – OpenClaw: 32 min, 1.2M tokens, $1.12 – Hermes: 2 h 14 min, 4.2M tokens, $6.42 Before the start we leveled the field: one model endpoint, equal step and token budgets, equal timeouts, full autonomy, memory wiped on all three. Atomic Agent reads images through its vision tool, so it interrogated the sketch 14 times until every room, door and window turned into data. Then it drafted the whole scene in its head six times, threw away five drafts, and wrote the finished 19.8 KB file in one single write. After that it opened Chrome, checked its own render, and only then replied. The only agent of the three that verified its work, and the only one that stopped on its own. OpenClaw was twice as fast and the cheapest of the three, but its image tool kept timing out mid-run, and it shipped the palest apartment of the day: white rooms, no floor colors, one texture visibly glitched, and furniture you have to squint to find. It read the full plan three times, cut 11 room crops, wrote the scene in chunks, and landed the fastest and cheapest apartment of the day in 32 minutes. Then it kept polishing the finished file until we pulled the plug. Hermes worked the longest: two hours, 97 model calls, 4.2M tokens, and the apartment came out wrong anyway: doors standing loose in the middle of rooms, a 2 by 1.8 bath sprawled across a quarter of the flat, furniture drifting away from the plan. It measured everything twice and still built the least accurate apartment. Atomic Agent will run Qwen3.8-27B locally on day zero, next week!
0:28

Qwen3.8-Max became the brain of Atomic Agent, Hermes and OpenClaw. We gave the same task: Turn a photo of a hand-drawn floor plan into an interactive 3D walkthrough of that apartment and open it in the browser. Outputs: – Atomic Agent: 66 min, 557K tokens, $2.01 – OpenClaw: 32 min, 1.2M tokens, $1.12 – Hermes: 2 h 14 min, 4.2M tokens, $6.42 Before the start we leveled the field: one model endpoint, equal step and token budgets, equal timeouts, full autonomy, memory wiped on all three. Atomic Agent reads images through its vision tool, so it interrogated the sketch 14 times until every room, door and window turned into data. Then it drafted the whole scene in its head six times, threw away five drafts, and wrote the finished 19.8 KB file in one single write. After that it opened Chrome, checked its own render, and only then replied. The only agent of the three that verified its work, and the only one that stopped on its own. OpenClaw was twice as fast and the cheapest of the three, but its image tool kept timing out mid-run, and it shipped the palest apartment of the day: white rooms, no floor colors, one texture visibly glitched, and furniture you have to squint to find. It read the full plan three times, cut 11 room crops, wrote the scene in chunks, and landed the fastest and cheapest apartment of the day in 32 minutes. Then it kept polishing the finished file until we pulled the plug. Hermes worked the longest: two hours, 97 model calls, 4.2M tokens, and the apartment came out wrong anyway: doors standing loose in the middle of rooms, a 2 by 1.8 bath sprawled across a quarter of the flat, furniture drifting away from the plan. It measured everything twice and still built the least accurate apartment. Atomic Agent will run Qwen3.8-27B locally on day zero, next week!

Atomic Agent

122,569 görüntüleme • 26 gün önce

Alibaba just released a coding model that hits 82 percent on SWE-Bench Verified. That is the highest score ever published for an open-source model. The weights are free. The license is Apache 2.0. You can run it today. The model is Qwen 4 Coder 32B. Here is what 82 percent on SWE-Bench Verified actually means. SWE-Bench Verified tests whether an AI can autonomously resolve real bugs pulled from real production GitHub repositories. Not synthetic exercises. Real open-source projects that real teams depend on. A model gets a bug report, reads the code, writes a fix, and either passes the test suite or it does not. At 82 percent, Qwen 4 Coder 32B resolves 82 out of every 100 real production bugs it is given. Without a human guiding it. On code it has never seen before. For comparison: Qwen 4 Coder 32B: 82 percent SWE-Bench Verified. Open source. Apache 2.0. Claude Fable 5: 80.3 percent SWE-Bench Pro. $10 input / $50 output per million tokens. Currently suspended. GPT-5.6 Sol: Competitive on Terminal-Bench. $5 input / $30 output per million tokens. An open-weight model that you can download and run for free just beat both of them on the benchmark designed to measure real software engineering capability. Here is the architecture. Qwen 4 Coder 32B is a 32 billion parameter dense model. Not a Mixture-of-Experts. Every parameter is active on every request. This matters for inference: a dense 32B model runs on 22 gigabytes of VRAM, which fits on a single high-end consumer GPU or a MacBook Pro with 64GB of unified memory. The smaller variant, Qwen 4 Coder 4B, runs at approximately 135 tokens per second on an M5 Max and fits inside 8 gigabytes of RAM. For a model with usable coding capability, that is a new bar for what fits in a single laptop. The training methodology continued Alibaba's approach of reinforcement learning on verifiable coding tasks. The model gets rewarded when its code passes tests. It gets penalized when it fails. Over millions of training steps, the model learns to write code that actually runs rather than code that looks plausible. License: Apache 2.0. Full commercial use. No attribution requirement. No revenue threshold. No monthly active user ceiling. Weights: Hugging Face, available today. Runs on: vLLM, Ollama, SGLang, and any standard GGUF-compatible inference engine. Qwen 4 32B also runs at approximately 135 tokens per second on an M5 Max chip, setting a new bar for what a sub-8GB model can do on Apple Silicon. The open-source coding model just beat the best closed-source model in the world on the benchmark designed to test whether AI can actually do software engineering. The weights are free. The subscription is optional. Source: Autom8Labs AI Insight July 2026, State of Open Source LLMs June 2026, Kunal Ganglani blog June 2026.
0:25

Alibaba just released a coding model that hits 82 percent on SWE-Bench Verified. That is the highest score ever published for an open-source model. The weights are free. The license is Apache 2.0. You can run it today. The model is Qwen 4 Coder 32B. Here is what 82 percent on SWE-Bench Verified actually means. SWE-Bench Verified tests whether an AI can autonomously resolve real bugs pulled from real production GitHub repositories. Not synthetic exercises. Real open-source projects that real teams depend on. A model gets a bug report, reads the code, writes a fix, and either passes the test suite or it does not. At 82 percent, Qwen 4 Coder 32B resolves 82 out of every 100 real production bugs it is given. Without a human guiding it. On code it has never seen before. For comparison: Qwen 4 Coder 32B: 82 percent SWE-Bench Verified. Open source. Apache 2.0. Claude Fable 5: 80.3 percent SWE-Bench Pro. $10 input / $50 output per million tokens. Currently suspended. GPT-5.6 Sol: Competitive on Terminal-Bench. $5 input / $30 output per million tokens. An open-weight model that you can download and run for free just beat both of them on the benchmark designed to measure real software engineering capability. Here is the architecture. Qwen 4 Coder 32B is a 32 billion parameter dense model. Not a Mixture-of-Experts. Every parameter is active on every request. This matters for inference: a dense 32B model runs on 22 gigabytes of VRAM, which fits on a single high-end consumer GPU or a MacBook Pro with 64GB of unified memory. The smaller variant, Qwen 4 Coder 4B, runs at approximately 135 tokens per second on an M5 Max and fits inside 8 gigabytes of RAM. For a model with usable coding capability, that is a new bar for what fits in a single laptop. The training methodology continued Alibaba's approach of reinforcement learning on verifiable coding tasks. The model gets rewarded when its code passes tests. It gets penalized when it fails. Over millions of training steps, the model learns to write code that actually runs rather than code that looks plausible. License: Apache 2.0. Full commercial use. No attribution requirement. No revenue threshold. No monthly active user ceiling. Weights: Hugging Face, available today. Runs on: vLLM, Ollama, SGLang, and any standard GGUF-compatible inference engine. Qwen 4 32B also runs at approximately 135 tokens per second on an M5 Max chip, setting a new bar for what a sub-8GB model can do on Apple Silicon. The open-source coding model just beat the best closed-source model in the world on the benchmark designed to test whether AI can actually do software engineering. The weights are free. The subscription is optional. Source: Autom8Labs AI Insight July 2026, State of Open Source LLMs June 2026, Kunal Ganglani blog June 2026.

Harman

41,278 görüntüleme • 1 ay önce

Agent Zero crushed Codex by 2.6× on token efficiency! We gave both agents the same task on the same local model (Qwen3.6 27B): build a retro Q*bert arcade game. Each agent wrote and rewrote it across iterations, adding enemies and a HUD. Outputs: • Agent Zero: 239K tokens, 53 min • Codex: 627K tokens, 51 min Both agents built a working Q*bert, but GitLawb 's Agent Zero is the cleaner build, with crisper textures and hops that land where they should. Codex refills the context window with its whole history every turn, so the model loses the thread and hallucinates code it never wrote. Agent Zero sends only what's new, so it keeps more of the window open and makes fewer mistakes. Run agents on local AI models in Atomic Chat!
0:25

Agent Zero crushed Codex by 2.6× on token efficiency! We gave both agents the same task on the same local model (Qwen3.6 27B): build a retro Q*bert arcade game. Each agent wrote and rewrote it across iterations, adding enemies and a HUD. Outputs: • Agent Zero: 239K tokens, 53 min • Codex: 627K tokens, 51 min Both agents built a working Q*bert, but GitLawb 's Agent Zero is the cleaner build, with crisper textures and hops that land where they should. Codex refills the context window with its whole history every turn, so the model loses the thread and hallucinates code it never wrote. Agent Zero sends only what's new, so it keeps more of the window open and makes fewer mistakes. Run agents on local AI models in Atomic Chat!

atomic.chat

320,187 görüntüleme • 1 ay önce

🧵 1/7 I gave a Flash model a written spec for a 3D physics game. It broke the task into steps, wrote the code, used the tools, ran the project, and eventually produced a billiards game I could actually play. Ant Ling Ling-3.0-flash has 124B total parameters, but activates just 5.1B per token. That low activation count didn’t limit it to a simple one-shot demo. It kept a multi-stage 3D build moving from spec to playable result. This test made one thing clear: stop using your biggest model for every step of an agent workflow. The project may be complex. Most of the execution steps inside it don’t need a heavyweight planner to rethink everything from scratch. Ling-3.0-flash is built to execute. Here’s the full hands-on video: 👇
1:30

🧵 1/7 I gave a Flash model a written spec for a 3D physics game. It broke the task into steps, wrote the code, used the tools, ran the project, and eventually produced a billiards game I could actually play. Ant Ling Ling-3.0-flash has 124B total parameters, but activates just 5.1B per token. That low activation count didn’t limit it to a simple one-shot demo. It kept a multi-stage 3D build moving from spec to playable result. This test made one thing clear: stop using your biggest model for every step of an agent workflow. The project may be complex. Most of the execution steps inside it don’t need a heavyweight planner to rethink everything from scratch. Ling-3.0-flash is built to execute. Here’s the full hands-on video: 👇

Lonely

121,029 görüntüleme • 1 ay önce

Qwen 3.7 Max looks incredible on paper. Then I spent $43 in 15 minutes of vibe coding. Best Chinese model on every benchmark. #1 on SWE-Bench Pro. #4 on BridgeBench UI. The scores are legit. But in a real workflow it came back with 15 errors on a single task. No subsidies like Claude Code or Codex. Cheaper per token, more expensive per task. Chinese models are getting closer. But "closer" still means you're burning more time and more money to get the same result GPT 5.5 one shots. Full review and BridgeBench breakdown below.
17:08

Qwen 3.7 Max looks incredible on paper. Then I spent $43 in 15 minutes of vibe coding. Best Chinese model on every benchmark. #1 on SWE-Bench Pro. #4 on BridgeBench UI. The scores are legit. But in a real workflow it came back with 15 errors on a single task. No subsidies like Claude Code or Codex. Cheaper per token, more expensive per task. Chinese models are getting closer. But "closer" still means you're burning more time and more money to get the same result GPT 5.5 one shots. Full review and BridgeBench breakdown below.

BridgeMind

15,246 görüntüleme • 3 ay önce

Qwen 3.8 Max beat Fable 5 at building 3D physics scenes for 7x cheaper! We gave two models the same task. Build three self-contained 3D scenes, each one HTML file with real physics that runs itself. Prompts: - A marble machine that lifts marbles up a wheel and drops them on a loop - A car factory assembly line - A sawmill cutting logs into planks Outputs: Qwen 3.8 Max: 46.6K tokens, $0.28 Fable 5: 38.7K tokens, $1.93 Qwen worked much harder on the details. Its textures and shadows look real. Fable 5 built a fast prototype that works and left it there. In the marble machine, Fable's marbles show up from nowhere at the top of the wheel. In Qwen's, you can see them go up. On the car line, Qwen's cars look much better and more real. Fable's sawmill is very strange. The saw looks wrong, and the log goes right through a beam before the saw cuts it. Atomic Chat will have day zero support to run Qwen 3.8 Max locally!
0:30

Sensitive content

Qwen 3.8 Max beat Fable 5 at building 3D physics scenes for 7x cheaper! We gave two models the same task. Build three self-contained 3D scenes, each one HTML file with real physics that runs itself. Prompts: - A marble machine that lifts marbles up a wheel and drops them on a loop - A car factory assembly line - A sawmill cutting logs into planks Outputs: Qwen 3.8 Max: 46.6K tokens, $0.28 Fable 5: 38.7K tokens, $1.93 Qwen worked much harder on the details. Its textures and shadows look real. Fable 5 built a fast prototype that works and left it there. In the marble machine, Fable's marbles show up from nowhere at the top of the wheel. In Qwen's, you can see them go up. On the car line, Qwen's cars look much better and more real. Fable's sawmill is very strange. The saw looks wrong, and the log goes right through a beam before the saw cuts it. Atomic Chat will have day zero support to run Qwen 3.8 Max locally!

atomic.chat

210,569 görüntüleme • 26 gün önce

Tally is launching an ICO. We believe we've built the best on-chain mechanism for raising capital via tokens — and we're going to prove it by using it ourselves. Over the next 60 days, we're documenting every step of what it looks like to launch a token the right way, in the United States, in 2026. We're demystifying a process that has historically be rife with abuse by bringing all the steps out into the open. Episode 1 👇
2:25

Tally is launching an ICO. We believe we've built the best on-chain mechanism for raising capital via tokens — and we're going to prove it by using it ourselves. Over the next 60 days, we're documenting every step of what it looks like to launch a token the right way, in the United States, in 2026. We're demystifying a process that has historically be rife with abuse by bringing all the steps out into the open. Episode 1 👇

Tally

66,207 görüntüleme • 6 ay önce

most agent memory is one file that grows forever and gets re-read every single turn. it overflows the context, overwrites old facts, loses the thread. that's why your agent feels sharp on day one and lost by week three. Sibyl Memory replaces the pile with a structure: → leaner context, lower token cost. a bloated agent file and flat memory get re-read in full every message. Sibyl Memory keeps a light hot layer and retrieves the rest on demand, so each turn carries only the tokens that matter. → relations are first-class. people link to projects, projects to deals, deals to the decisions behind them. ask about one partner and the agent surfaces the whole connected web around them. that relational context is what lets it help run a company. → one source of truth per fact, person, and project. no duplicates, no silent overwrites. works with hermes, claude code, and codex. beta open.
0:34

most agent memory is one file that grows forever and gets re-read every single turn. it overflows the context, overwrites old facts, loses the thread. that's why your agent feels sharp on day one and lost by week three. Sibyl Memory replaces the pile with a structure: → leaner context, lower token cost. a bloated agent file and flat memory get re-read in full every message. Sibyl Memory keeps a light hot layer and retrieves the rest on demand, so each turn carries only the tokens that matter. → relations are first-class. people link to projects, projects to deals, deals to the decisions behind them. ask about one partner and the agent surfaces the whole connected web around them. that relational context is what lets it help run a company. → one source of truth per fact, person, and project. no duplicates, no silent overwrites. works with hermes, claude code, and codex. beta open.

SIBYL

83,832 görüntüleme • 3 ay önce

We are excited to share that Logan Kilpatrick joined us on The Bench to discuss Google's new Gemini 3.5 Flash: why it's deliberately more persistent and capable than previous Flash models, how it hit #1 on our FinanceAgent Benchmark taking 82 steps where competitors stopped at 13, and what justifies the price increase. We also get into why AI benchmarks need a paradigm shift, the trade-off of building everything vs staying focused, the Pope, and why Omni might kill the Subway Surfers content era. 0:11:00 – Flash is being rebased for the agent era, not just a cheaper model anymore 0:14:03 – Persistence by design: 82 tool calls vs competitors' 13 0:17:52 – Why pricing went up and how Google thinks about value per token 0:22:55 – Coding performance: from 20th to 10th place in one generation 0:28:28 – Why benchmarks have historically been misleading and what the new era of evaluation looks like 0:29:28 Logan on why Google has the best researchers in the world 0:36:16 – The cost of being Google 0:39:07 – The Pope’s encyclical on AI and whether most people see frontier intelligence as a good thing 0:51:12 – Why Omni is the thing that recently clicked
54:53

We are excited to share that Logan Kilpatrick joined us on The Bench to discuss Google's new Gemini 3.5 Flash: why it's deliberately more persistent and capable than previous Flash models, how it hit #1 on our FinanceAgent Benchmark taking 82 steps where competitors stopped at 13, and what justifies the price increase. We also get into why AI benchmarks need a paradigm shift, the trade-off of building everything vs staying focused, the Pope, and why Omni might kill the Subway Surfers content era. 0:11:00 – Flash is being rebased for the agent era, not just a cheaper model anymore 0:14:03 – Persistence by design: 82 tool calls vs competitors' 13 0:17:52 – Why pricing went up and how Google thinks about value per token 0:22:55 – Coding performance: from 20th to 10th place in one generation 0:28:28 – Why benchmarks have historically been misleading and what the new era of evaluation looks like 0:29:28 Logan on why Google has the best researchers in the world 0:36:16 – The cost of being Google 0:39:07 – The Pope’s encyclical on AI and whether most people see frontier intelligence as a good thing 0:51:12 – Why Omni is the thing that recently clicked

Vals AI

16,486 görüntüleme • 3 ay önce

GPT-5.6 LUNA AND DEEPSEEK V4 FLASH ARE BOTH FREE RIGHT NOW -- ONE IN A CHATBOT, THE OTHER IN A BROWSER IDE • ChatGPT, the chatbot > GPT-5.6 Luna is now the default on Free and Go: > Unlimited text chats and a Think button land next week. • MonkeyCode, the browser IDE > 30M tokens a day, resets daily, no card: > DeepSeek V4 Flash, Qwen, GLM, Kimi and MiniMax behind it. > Cloud workspace at 1 vCPU and 4GB, nothing to install. Luna posts 62% fewer factual errors than the model it replaces. MonkeyCode is open source from Chaitin and runs on their cloud, so aim it at side projects. One is live today -> The other arrives next week.
0:24

GPT-5.6 LUNA AND DEEPSEEK V4 FLASH ARE BOTH FREE RIGHT NOW -- ONE IN A CHATBOT, THE OTHER IN A BROWSER IDE • ChatGPT, the chatbot > GPT-5.6 Luna is now the default on Free and Go: > Unlimited text chats and a Think button land next week. • MonkeyCode, the browser IDE > 30M tokens a day, resets daily, no card: > DeepSeek V4 Flash, Qwen, GLM, Kimi and MiniMax behind it. > Cloud workspace at 1 vCPU and 4GB, nothing to install. Luna posts 62% fewer factual errors than the model it replaces. MonkeyCode is open source from Chaitin and runs on their cloud, so aim it at side projects. One is live today -> The other arrives next week.

slash1s

33,804 görüntüleme • 21 gün önce

Robinhood chain is growing fast, and every new, growing chain attracts rugs. Few days ago, I was shilled a token by a fren & I bought it few minutes after, it was up 3x. $700 should be $2.1k Checked my token balance & it was 0. Checked my tx & still not showing Apparently, scammer dev puts the tax at 99% and lets the contract transfer tokens after you bought. That's how I got rugged of $700 Many other ways people have gotten rugged on Robinhood in similar & other ways Luckily, we have Hood Terminal to the rescue Hood Terminal has built a feature where you get to scan tokens before apeing. You get a token score and the tool tells you if there's danger or not. It also has an anti-rug system designed to keep you safe on your apes. Apart from the risk intelligence tool, they've also teased other onchain analytics tools are on the way. Check them out and stop getting rugged Cheers to make 100x onchain!
0:37

Robinhood chain is growing fast, and every new, growing chain attracts rugs. Few days ago, I was shilled a token by a fren & I bought it few minutes after, it was up 3x. $700 should be $2.1k Checked my token balance & it was 0. Checked my tx & still not showing Apparently, scammer dev puts the tax at 99% and lets the contract transfer tokens after you bought. That's how I got rugged of $700 Many other ways people have gotten rugged on Robinhood in similar & other ways Luckily, we have Hood Terminal to the rescue Hood Terminal has built a feature where you get to scan tokens before apeing. You get a token score and the tool tells you if there's danger or not. It also has an anti-rug system designed to keep you safe on your apes. Apart from the risk intelligence tool, they've also teased other onchain analytics tools are on the way. Check them out and stop getting rugged Cheers to make 100x onchain!

InfoSpace OG

11,318 görüntüleme • 1 ay önce

The most downloaded AI on earth is now Chinese. Alibaba just gave away a model that matches Claude's flagship, and it literally runs on a $700 used graphics card. The Qwen models crossed 3 BILLION downloads in six months. Hugging Face counted 418 million downloads for Google this year, and 227 million for Meta. Alibaba cleared more than four times both of them combined. Then today it released Qwen3.8-27B under an Apache 2.0 license. The model has 27 billion parameters, native vision, and a 262,000 token context window. Developers are running it locally on 17 gigabytes of memory, on used cards that cost a few hundred dollars. Alibaba's own benchmark table claims it beats Opus 4.6 Max on computer use by 84.3 to 72.7, on mobile use by 81.9 to 62, and on visual math by 94.6 to 65.5. Those numbers come from the vendor and nobody has independently verified them yet, so treat them as a claim. But the generation over generation jumps are harder to wave away: On DeepSWE the score went from 13.3 to 42.2. On software engineering it went from 49.3 to 79.0. That happened in ONE release cycle. And Apache 2.0 means anyone can download the weights, modify them, build products on them, sell those products, and never pay or ask permission. It cannot be revoked. Once the file is on your drive it is yours permanently. 3 billion downloads means those files already sit on machines in every country on Earth. Alibaba could delete everything tomorrow and it would change nothing. Washington spent 4 years building an export control regime around chips, model weights, and entity lists. Every piece of it assumes a chokepoint exists somewhere. A fab, a shipment, a company that can be told no. But there is no chokepoint for a file that has already been copied three billion times. And the copying compounds. Hugging Face counted 151,448 models built on top of Qwen, which is 2.6x Meta's entire footprint and 4.7x the number of Llama repositories. New ones appear at roughly 200 a day. The report says Qwen has become "part of the default workflow for developers deciding what models to fine-tune and deploy." Alibaba is also pushing Qwen through its cloud into Southeast Asia and Africa, markets where American labs have almost no presence, and where a very large share of the next generation of developers will learn to build. Meta and Nvidia have both rushed out new open models in recent weeks. That is what a response looks like when you feel the floor move. And to be clear, these are download and derivative numbers, not usage numbers. ChatGPT and Claude cannot be downloaded at all, so they do not appear in this comparison. What the figures measure is what developers choose to build on top of, which is a different question from what consumers type into a box. That is also why it matters MORE. Consumer habits change in an afternoon. Infrastructure choices last a decade, because everything built on top has to be rewritten to undo them. The American labs are valued on an assumption that frontier intelligence stays scarce, expensive, and rented by the token. Alibaba just made a version of it free, permanent, and small enough to run on hardware people already own. You will not get an announcement when the software you use every day starts running on a Chinese model underneath. Go and count how many of the tools you rely on could be rebuilt on free weights this year.
0:39

The most downloaded AI on earth is now Chinese. Alibaba just gave away a model that matches Claude's flagship, and it literally runs on a $700 used graphics card. The Qwen models crossed 3 BILLION downloads in six months. Hugging Face counted 418 million downloads for Google this year, and 227 million for Meta. Alibaba cleared more than four times both of them combined. Then today it released Qwen3.8-27B under an Apache 2.0 license. The model has 27 billion parameters, native vision, and a 262,000 token context window. Developers are running it locally on 17 gigabytes of memory, on used cards that cost a few hundred dollars. Alibaba's own benchmark table claims it beats Opus 4.6 Max on computer use by 84.3 to 72.7, on mobile use by 81.9 to 62, and on visual math by 94.6 to 65.5. Those numbers come from the vendor and nobody has independently verified them yet, so treat them as a claim. But the generation over generation jumps are harder to wave away: On DeepSWE the score went from 13.3 to 42.2. On software engineering it went from 49.3 to 79.0. That happened in ONE release cycle. And Apache 2.0 means anyone can download the weights, modify them, build products on them, sell those products, and never pay or ask permission. It cannot be revoked. Once the file is on your drive it is yours permanently. 3 billion downloads means those files already sit on machines in every country on Earth. Alibaba could delete everything tomorrow and it would change nothing. Washington spent 4 years building an export control regime around chips, model weights, and entity lists. Every piece of it assumes a chokepoint exists somewhere. A fab, a shipment, a company that can be told no. But there is no chokepoint for a file that has already been copied three billion times. And the copying compounds. Hugging Face counted 151,448 models built on top of Qwen, which is 2.6x Meta's entire footprint and 4.7x the number of Llama repositories. New ones appear at roughly 200 a day. The report says Qwen has become "part of the default workflow for developers deciding what models to fine-tune and deploy." Alibaba is also pushing Qwen through its cloud into Southeast Asia and Africa, markets where American labs have almost no presence, and where a very large share of the next generation of developers will learn to build. Meta and Nvidia have both rushed out new open models in recent weeks. That is what a response looks like when you feel the floor move. And to be clear, these are download and derivative numbers, not usage numbers. ChatGPT and Claude cannot be downloaded at all, so they do not appear in this comparison. What the figures measure is what developers choose to build on top of, which is a different question from what consumers type into a box. That is also why it matters MORE. Consumer habits change in an afternoon. Infrastructure choices last a decade, because everything built on top has to be rewritten to undo them. The American labs are valued on an assumption that frontier intelligence stays scarce, expensive, and rented by the token. Alibaba just made a version of it free, permanent, and small enough to run on hardware people already own. You will not get an announcement when the software you use every day starts running on a Chinese model underneath. Go and count how many of the tools you rely on could be rebuilt on free weights this year.

Ricardo

80,611 görüntüleme • 13 gün önce

SITUATION EXPLAINED: A stealth model on OpenRouter is beating Fable and Sol on coding, and nobody knows who made it. • Ox Alpha has a 1M token context window, text, image, and video input, free for a week, with capacity for 100 trillion tokens a day • On a DeepSWE subset it hit 80%, against Fable's 65%, GLM-5.3's 62%, and Sol's 52% • The tokenizer is one-for-one identical to GLM's, and a live test asking whether Taiwan is part of China returned a nearly identical answer • This is the 5th anonymous drop in six months, and the previous four were all claimed by Chinese labs: Zhipu, Xiaomi, Ant, and Meituan Theo Jaffee: " There's no way it's a Flash model. If that's a Flash model, then holy shit, the entire United States is cooked. If that's a Flash model, we need to short the stock market now. Not financial advice."
15:43

SITUATION EXPLAINED: A stealth model on OpenRouter is beating Fable and Sol on coding, and nobody knows who made it. • Ox Alpha has a 1M token context window, text, image, and video input, free for a week, with capacity for 100 trillion tokens a day • On a DeepSWE subset it hit 80%, against Fable's 65%, GLM-5.3's 62%, and Sol's 52% • The tokenizer is one-for-one identical to GLM's, and a live test asking whether Taiwan is part of China returned a nearly identical answer • This is the 5th anonymous drop in six months, and the previous four were all claimed by Chinese labs: Zhipu, Xiaomi, Ant, and Meituan Theo Jaffee: " There's no way it's a Flash model. If that's a Flash model, then holy shit, the entire United States is cooked. If that's a Flash model, we need to short the stock market now. Not financial advice."

MTS

376,143 görüntüleme • 8 gün önce

Anthropic's in trouble, again. The entire Claude experience is now available at 1/6th the price. Kimi now does everything Claude does, powered by K2.6, a 1-trillion-parameter MoE model that activates only 32B parameters per token. It covers all three features Claude has (Chat, Code, and Cowork): 1) Kimi Chat runs in four modes - Instant for fast responses - Thinking for deep reasoning - Agent for multi-step execution - and Agent Swarm for parallel workloads. There's a 262K context window across all of them. 2) Kimi Code is the open-source CLI coding agent with K2.6 as the default backend. K2.6 ranked #1 on OpenRouter's programming leaderboard by weekly usage. 3) Kimi Agent is the Cowork equivalent. It generates: - full websites with database and auth - presentation decks (editable PPTX output) - spreadsheets with formulas and charts - word docs and structured research reports. On top of this, Kimi K2.6 is also trained to decompose tasks into up to 300 parallel sub-agents. This helps it retain coherence even across 4,000+ tool calls in a single run, with sessions sustaining up to 13 hours. On SWE-Bench Pro: - Kimi K2.6 → 58.6 - GPT-5.4 xhigh → 57.7 - Gemini 3.1 Pro → 54.2 - Claude Opus 4.6 → 53.4 Kimi K2.6 model is open weights and self-hostable on 4x H100s in INT4. Find the link to the HuggingFace model page in the replies!
0:41

Anthropic's in trouble, again. The entire Claude experience is now available at 1/6th the price. Kimi now does everything Claude does, powered by K2.6, a 1-trillion-parameter MoE model that activates only 32B parameters per token. It covers all three features Claude has (Chat, Code, and Cowork): 1) Kimi Chat runs in four modes - Instant for fast responses - Thinking for deep reasoning - Agent for multi-step execution - and Agent Swarm for parallel workloads. There's a 262K context window across all of them. 2) Kimi Code is the open-source CLI coding agent with K2.6 as the default backend. K2.6 ranked #1 on OpenRouter's programming leaderboard by weekly usage. 3) Kimi Agent is the Cowork equivalent. It generates: - full websites with database and auth - presentation decks (editable PPTX output) - spreadsheets with formulas and charts - word docs and structured research reports. On top of this, Kimi K2.6 is also trained to decompose tasks into up to 300 parallel sub-agents. This helps it retain coherence even across 4,000+ tool calls in a single run, with sessions sustaining up to 13 hours. On SWE-Bench Pro: - Kimi K2.6 → 58.6 - GPT-5.4 xhigh → 57.7 - Gemini 3.1 Pro → 54.2 - Claude Opus 4.6 → 53.4 Kimi K2.6 model is open weights and self-hostable on 4x H100s in INT4. Find the link to the HuggingFace model page in the replies!

Avi Chawla

110,057 görüntüleme • 3 ay önce

i gave a 9B model an agentic OS and told it to build Octopus Invaders from scratch. Qwen 3.5 9B running through Hermes Agent with full tool access on a single 3060 12GB. browser, file system, code execution, terminal. it reads the prompt, plans the architecture, writes every file, and serves the game. one prompt, zero steering, 11 minutes. i've been going back and forth with this model for days now. text, code, reasoning. it handles small chats, batching, bash scripting. at 9B it's fast enough to feel interactive and genuinely useful for local workflows where you'd normally reach for an API. but this is the real test. can it architect and build a full game autonomously without a single correction? it's coding right now. results coming next.
11:40

i gave a 9B model an agentic OS and told it to build Octopus Invaders from scratch. Qwen 3.5 9B running through Hermes Agent with full tool access on a single 3060 12GB. browser, file system, code execution, terminal. it reads the prompt, plans the architecture, writes every file, and serves the game. one prompt, zero steering, 11 minutes. i've been going back and forth with this model for days now. text, code, reasoning. it handles small chats, batching, bash scripting. at 9B it's fast enough to feel interactive and genuinely useful for local workflows where you'd normally reach for an API. but this is the real test. can it architect and build a full game autonomously without a single correction? it's coding right now. results coming next.

Sudo su

65,570 görüntüleme • 5 ay önce

Atomic Agent beat Hermes on GAIA: 69.8% vs 58.5%, and it was 1.6x faster! We ran both agents through the full GAIA Level 1 benchmark, 53 real-world tasks, same 4-bit qwen-3.6-35b on the same Apple M4 Max. Results: ✦ Atomic Agent: 37 of 53 solved, done in 3h 12m ✦ Hermes Agent: 31 of 53 solved, took 5h 10m Atomic solved 6 more tasks and finished nearly 2 hours sooner. Hermes ran into the 900s timeout on 7 tasks; Atomic on just 2. Hermes burned 71% of its total time on tasks it still failed, Atomic, 48%. Where it showed: ✦ Audre Lorde poem, which stanza is indented: Atomic pushed through a dead source, switched tools, and answered in 7.6 min. Hermes ran the full clock and returned a blank. ✦ Vietnamese specimens, which city they ended up in: Atomic pulled it from the first source and normalized the answer in 33s. Hermes spent 7.3 min and never answered. ✦ The dinosaur featured-article nominator: Atomic walked the Wikipedia chain to "FunkMonk" in 57s. Hermes guessed a wrong name after 11 min. Atomic keeps a byte-stable prompt prefix, so llama-server reuses the KV-cache instead of re-encoding the whole context every turn, and it emits one JSON array of tool calls per inference, then compresses results back instead of pasting them in full, so the context never balloons and a small model stays sharp deep into a task. On top of that a no-progress guard vetoes repeated identical tool calls (warn at 3, hard veto at 5) and forces a reply, so Atomic never sinks 15 minutes into re-scanning one page the way Hermes did. Both agents missed some of the same questions, and on a few Hermes got there and Atomic did not, usually format slips where Atomic computed the right number but printed the working instead of the bare value. But on identical hardware and identical weights, the runtime that reuses its cache and refuses to spin came out ahead on accuracy and speed. Getting this from the runtime alone is wild. Run the same 53 GAIA tasks on Atomic Agent!
0:27

Atomic Agent beat Hermes on GAIA: 69.8% vs 58.5%, and it was 1.6x faster! We ran both agents through the full GAIA Level 1 benchmark, 53 real-world tasks, same 4-bit qwen-3.6-35b on the same Apple M4 Max. Results: ✦ Atomic Agent: 37 of 53 solved, done in 3h 12m ✦ Hermes Agent: 31 of 53 solved, took 5h 10m Atomic solved 6 more tasks and finished nearly 2 hours sooner. Hermes ran into the 900s timeout on 7 tasks; Atomic on just 2. Hermes burned 71% of its total time on tasks it still failed, Atomic, 48%. Where it showed: ✦ Audre Lorde poem, which stanza is indented: Atomic pushed through a dead source, switched tools, and answered in 7.6 min. Hermes ran the full clock and returned a blank. ✦ Vietnamese specimens, which city they ended up in: Atomic pulled it from the first source and normalized the answer in 33s. Hermes spent 7.3 min and never answered. ✦ The dinosaur featured-article nominator: Atomic walked the Wikipedia chain to "FunkMonk" in 57s. Hermes guessed a wrong name after 11 min. Atomic keeps a byte-stable prompt prefix, so llama-server reuses the KV-cache instead of re-encoding the whole context every turn, and it emits one JSON array of tool calls per inference, then compresses results back instead of pasting them in full, so the context never balloons and a small model stays sharp deep into a task. On top of that a no-progress guard vetoes repeated identical tool calls (warn at 3, hard veto at 5) and forces a reply, so Atomic never sinks 15 minutes into re-scanning one page the way Hermes did. Both agents missed some of the same questions, and on a few Hermes got there and Atomic did not, usually format slips where Atomic computed the right number but printed the working instead of the bare value. But on identical hardware and identical weights, the runtime that reuses its cache and refuses to spin came out ahead on accuracy and speed. Getting this from the runtime alone is wild. Run the same 53 GAIA tasks on Atomic Agent!

Atomic Agent

111,357 görüntüleme • 1 ay önce

A $351 MINI PC IS RUNNING 26-BILLION-PARAMETER AI MODELS AT 20 TOKENS/SEC AND HERMES AGENT ON TOP OF IT This is the Minisforum UM790 Pro. AMD Ryzen 9 7940HS, Radeon 780M iGPU, 48GB DDR5. The BIOS reports the GPU has 4GB of VRAM Here's the part people get wrong. The 780M has no dedicated VRAM at all it borrows from system RAM. Vulkan ignores the BIOS number and reads the full 48GB pool directly That's the whole trick. 21+ GB allocated to model weights on a "4GB" GPU. Because the models are MoE, only 4 billion of the parameters activate per token a fraction of the reads a dense model needs. That's why 20 tok/s works here Gemma 4 26B MoE holds 19.5 tok/s with 196K context. Qwen3.5-35B-A3B holds 20.8. Nemotron Cascade 2 clears 24.8. A dense 31B, by contrast, drops to 4 tok/s it reads the entire model every token, no way around it On top: Hermes Agent, full agentic workflows terminal, file ops, web, 40+ tools against local models only. No API keys. The wall between you and a usable local agent used to be a GPU you couldn't afford. Now it's a BIOS setting most people never check Bookmark this & Try it yourself ↓
1:22

A $351 MINI PC IS RUNNING 26-BILLION-PARAMETER AI MODELS AT 20 TOKENS/SEC AND HERMES AGENT ON TOP OF IT This is the Minisforum UM790 Pro. AMD Ryzen 9 7940HS, Radeon 780M iGPU, 48GB DDR5. The BIOS reports the GPU has 4GB of VRAM Here's the part people get wrong. The 780M has no dedicated VRAM at all it borrows from system RAM. Vulkan ignores the BIOS number and reads the full 48GB pool directly That's the whole trick. 21+ GB allocated to model weights on a "4GB" GPU. Because the models are MoE, only 4 billion of the parameters activate per token a fraction of the reads a dense model needs. That's why 20 tok/s works here Gemma 4 26B MoE holds 19.5 tok/s with 196K context. Qwen3.5-35B-A3B holds 20.8. Nemotron Cascade 2 clears 24.8. A dense 31B, by contrast, drops to 4 tok/s it reads the entire model every token, no way around it On top: Hermes Agent, full agentic workflows terminal, file ops, web, 40+ tools against local models only. No API keys. The wall between you and a usable local agent used to be a GPU you couldn't afford. Now it's a BIOS setting most people never check Bookmark this & Try it yourself ↓

slash1s

35,742 görüntüleme • 2 ay önce

So comes the biggest threat to the AI trade, which relies on ever-increasing revenues to the AI model companies to support ever-increasing valuations. Those revenues are a function of price and quantity. Quantities are all good, from my perspective. Despite increasing competition from open source and Chinese models, OpenAI and Anthropic continue adding more businesses and drawing more token usage from new and existing customers. Price, as a lever, is deteriorating. Over the last month, the average cost per million tokens has fallen 27%. This week, OpenAI cut prices for it's frontier model by another 20%. Further gains to the model companies will have to rely on driving more volume, and driving it even faster as prices fall, to support their valuations.
0:44

So comes the biggest threat to the AI trade, which relies on ever-increasing revenues to the AI model companies to support ever-increasing valuations. Those revenues are a function of price and quantity. Quantities are all good, from my perspective. Despite increasing competition from open source and Chinese models, OpenAI and Anthropic continue adding more businesses and drawing more token usage from new and existing customers. Price, as a lever, is deteriorating. Over the last month, the average cost per million tokens has fallen 27%. This week, OpenAI cut prices for it's frontier model by another 20%. Further gains to the model companies will have to rely on driving more volume, and driving it even faster as prices fall, to support their valuations.

Ara Kharazian

34,275 görüntüleme • 5 gün önce

If you are running local LLMs without N-gram speculative decoding, you are wasting massive amounts of compute. Whether your AI is editing a document, outputting structured JSON, or rewriting boilerplate templates, a huge chunk of the text it generates is highly repetitive or already exists right there in the prompt. Standard decoding wastes expensive GPU compute cycles "re thinking" every single token. By adding one hidden flag in llama.cpp, you can instantly fast forward through the repetition. Zero draft models. Zero extra VRAM. And virtually zero compute overhead. Google Colab hands you an enterprise grade NVIDIA Tesla T4 GPU with 16GB of VRAM for free. It’s the perfect Ubuntu Linux sandbox to build a bleeding edge inference engine from scratch. Recently, I showed you how to double your local speeds using MTP (Multi Token Prediction). But MTP requires a secondary neural network draft model. That eats into your precious VRAM (slightly though) and burns extra compute for every guess it makes. N-gram Speculative Decoding gives you a massive speed boost for exactly 0 memory cost and minimal compute. And it's faster than MTP when it works. Here is how it actually works under the hood: Standard autoregressive decoding is slow because it predicts one token at a time. If you ask an agent to format a long JSON object or update one line in an HTML file, it runs heavy matrix multiplications to calculate the probability of every single bracket, space, and letter from scratch. N-gram changes the game. It acts as a lightweight caching system. Instead of running heavy neural network math to guess the next word, it uses a simple hash table. Whenever the LLM starts outputting a sequence of tokens that already exists anywhere in its context window, N-gram instantly recognizes the pattern. Because it is just doing lightning fast string matching, the compute cost is practically zero. It "fast forwards" through the text, drafting the boilerplate instantly from memory, and the main model just verifies it in parallel. Pure speed. Using quantized GGUFs from Unsloth via HuggingFace, I spun up DeepMind’s massive Gemma 4 26B A4B QAT MoE on a free Colab instance to test this. Just look at the raw benchmark data on code editing task: Without N-gram: [ Prompt: 638.6 t/s | Generation: 45.9 t/s ] With N-gram: [ Prompt: 601.9 t/s | Generation: 107.1 t/s ] Here is the exact llama.cpp CLI command to activate it. Notice we don't even need the --model-draft flag: ./llama-cli -m gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf -cnv -n 6000 -c 12000 -ngl 99 -fa on --spec-type ngram-mod Stop waiting for your GPU to re calculate words it already knows. I’ve built a free, interactive, cell by cell Google Colab notebook that lets you test this live in your browser. You can literally chat with the model and watch the text generation speed absolutely fly on the second turn when you ask it to edit a file. There are additional parameters for ngram-mod that you can tune once you get it working with the single flag. Link to the free Colab Notebook is in the comments below. It walks you through the entire stack: pulling pre built llama.cpp CUDA binaries for Linux, fetching GGUFs from HuggingFace, and spinning up the inference engine with ngram-mod from scratch. Let me know if you have already tried ngram-mod
0:23

If you are running local LLMs without N-gram speculative decoding, you are wasting massive amounts of compute. Whether your AI is editing a document, outputting structured JSON, or rewriting boilerplate templates, a huge chunk of the text it generates is highly repetitive or already exists right there in the prompt. Standard decoding wastes expensive GPU compute cycles "re thinking" every single token. By adding one hidden flag in llama.cpp, you can instantly fast forward through the repetition. Zero draft models. Zero extra VRAM. And virtually zero compute overhead. Google Colab hands you an enterprise grade NVIDIA Tesla T4 GPU with 16GB of VRAM for free. It’s the perfect Ubuntu Linux sandbox to build a bleeding edge inference engine from scratch. Recently, I showed you how to double your local speeds using MTP (Multi Token Prediction). But MTP requires a secondary neural network draft model. That eats into your precious VRAM (slightly though) and burns extra compute for every guess it makes. N-gram Speculative Decoding gives you a massive speed boost for exactly 0 memory cost and minimal compute. And it's faster than MTP when it works. Here is how it actually works under the hood: Standard autoregressive decoding is slow because it predicts one token at a time. If you ask an agent to format a long JSON object or update one line in an HTML file, it runs heavy matrix multiplications to calculate the probability of every single bracket, space, and letter from scratch. N-gram changes the game. It acts as a lightweight caching system. Instead of running heavy neural network math to guess the next word, it uses a simple hash table. Whenever the LLM starts outputting a sequence of tokens that already exists anywhere in its context window, N-gram instantly recognizes the pattern. Because it is just doing lightning fast string matching, the compute cost is practically zero. It "fast forwards" through the text, drafting the boilerplate instantly from memory, and the main model just verifies it in parallel. Pure speed. Using quantized GGUFs from Unsloth via HuggingFace, I spun up DeepMind’s massive Gemma 4 26B A4B QAT MoE on a free Colab instance to test this. Just look at the raw benchmark data on code editing task: Without N-gram: [ Prompt: 638.6 t/s | Generation: 45.9 t/s ] With N-gram: [ Prompt: 601.9 t/s | Generation: 107.1 t/s ] Here is the exact llama.cpp CLI command to activate it. Notice we don't even need the --model-draft flag: ./llama-cli -m gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf -cnv -n 6000 -c 12000 -ngl 99 -fa on --spec-type ngram-mod Stop waiting for your GPU to re calculate words it already knows. I’ve built a free, interactive, cell by cell Google Colab notebook that lets you test this live in your browser. You can literally chat with the model and watch the text generation speed absolutely fly on the second turn when you ask it to edit a file. There are additional parameters for ngram-mod that you can tune once you get it working with the single flag. Link to the free Colab Notebook is in the comments below. It walks you through the entire stack: pulling pre built llama.cpp CUDA binaries for Linux, fetching GGUFs from HuggingFace, and spinning up the inference engine with ngram-mod from scratch. Let me know if you have already tried ngram-mod

Alok

31,765 görüntüleme • 1 ay önce

Anthropic shipped 125 settings for Claude The official docs cover 40 One developer found the other 85 and his API bill dropped from $340 to $87 - not by using a cheaper model - not by writing shorter prompts just by moving one line in a config file to the right place > memory scoped per project → past clients never bleed into new work > Extended Thinking on Light by default → 18–25% fewer Opus tokens in week one > cache_control moved to the right line → the fix that turned a $340 bill into $87 > plugins and MCP servers toggled off when idle → saved 25–40K tokens per session > per-project model override → Haiku for docs, Sonnet for infra, Opus only where it matters most Claude users are running a $100/month tool at 30% of its actual capability
0:49

Anthropic shipped 125 settings for Claude The official docs cover 40 One developer found the other 85 and his API bill dropped from $340 to $87 - not by using a cheaper model - not by writing shorter prompts just by moving one line in a config file to the right place > memory scoped per project → past clients never bleed into new work > Extended Thinking on Light by default → 18–25% fewer Opus tokens in week one > cache_control moved to the right line → the fix that turned a $340 bill into $87 > plugins and MCP servers toggled off when idle → saved 25–40K tokens per session > per-project model override → Haiku for docs, Sonnet for infra, Opus only where it matters most Claude users are running a $100/month tool at 30% of its actual capability

Shadow Nick

380,211 görüntüleme • 3 ay önce