←
Cruz3,943 posts
Cruz avatar

Cruz

@ViC305

God-fearing husband & father. AI Engineer. Fine-tuning + day-zero EXL3 & GGUF quants + benchmarks • Local LLMs • honest tok/s • fitting big models on small GPUs

◷ Joined June 2009▧ 507 Photos and videos
Cruz avatar
Cruz@ViC305·Oct 20

🚨 𝗜 𝗝𝗨𝗦𝗧 𝗔𝗖𝗛𝗜𝗘𝗩𝗘𝗗 𝗦𝗢𝗠𝗘𝗧𝗛𝗜𝗡𝗚 𝗜𝗡𝗖𝗥𝗘𝗗𝗜𝗕𝗟𝗘 𝗢𝗡 𝗢𝗡𝗘 𝗗𝗚𝗫 𝗦𝗣𝗔𝗥𝗞! 🧪🔥 With the support of @AWERXAI, I'm excited to release something VERY special for the local AI community! 🧠 @deepseek_ai 𝗗𝗲𝗲𝗽𝗦𝗲𝗲𝗸 𝗩𝟰.𝟭 𝗙𝗹𝗮𝘀𝗵 𝟭𝟬𝟭𝟳 📦 𝗦𝗔𝗚𝗘-𝗘𝗫𝗟𝟯 𝗮𝘁 𝗷𝘂𝘀𝘁 𝟭.𝟲𝟮 𝗯𝗽𝘄 𝗲𝘅𝗽𝗲𝗿𝘁𝘀 💻 𝟵𝟵.𝟲𝟭% 𝗖𝗢𝗗𝗘 𝗮𝗻𝘀𝘄𝗲𝗿-𝘁𝗼𝗸𝗲𝗻 𝗮𝗴𝗿𝗲𝗲𝗺𝗲𝗻𝘁! ⚡ 𝟯𝟱.𝟱 𝘁𝗼𝗸/𝘀 𝗱𝗲𝗰𝗼𝗱𝗲 𝗼𝗻 𝘃𝗟𝗟𝗠! 🚀 𝟭,𝟭𝟴𝟵 𝘁𝗼𝗸/𝘀 𝗠𝗔𝗫 𝗣𝗥𝗘𝗙𝗜𝗟𝗟! 🧠 𝟮𝟮𝟵𝗞 𝗖𝗧𝗫 (𝘀𝘁𝗶𝗹𝗹 𝘄𝗼𝗿𝗸𝗶𝗻𝗴 𝘁𝗼 𝗽𝘂𝘀𝗵 𝗶𝘁 𝗳𝘂𝗿𝘁𝗵𝗲𝗿!) 𝗔𝗟𝗟 𝗢𝗡 𝗢𝗡𝗘. 𝗗𝗚𝗫. 𝗦𝗣𝗔𝗥𝗞. 🤯 And YES, the quantized weights are AVAILABLE NOW! 💚 ━━━━━━━━━━━━━━━━━━━━ 🧪 𝗧𝗛𝗘 𝗟𝗔𝗦𝗧 𝟳𝟮 𝗛𝗢𝗨𝗥𝗦 𝗛𝗔𝗩𝗘 𝗕𝗘𝗘𝗡 𝗜𝗡𝗦𝗔𝗡𝗘! Just a few days ago, I published my article about @nvidiaai's FP4-Local format: The Good, The Bad & The Ugly. One thing really stuck with me... NVIDIA demonstrated what was possible when aggressive quantization and intelligent optimization come together. But there's a problem. 𝗥𝗘𝗣𝗥𝗢𝗗𝗨𝗖𝗜𝗡𝗚 𝗠𝗔𝗦𝗦𝗜𝗩𝗘 𝗤𝗨𝗔𝗡𝗧𝗜𝗭𝗔𝗧𝗜𝗢𝗡 𝗧𝗥𝗔𝗜𝗡𝗜𝗡𝗚 𝗣𝗜𝗣𝗘𝗟𝗜𝗡𝗘𝗦 𝗜𝗦 𝗢𝗨𝗧 𝗢𝗙 𝗥𝗘𝗔𝗖𝗛 𝗙𝗢𝗥 𝗠𝗢𝗦𝗧 𝗟𝗢𝗖𝗔𝗟 𝗔𝗜 𝗘𝗡𝗧𝗛𝗨𝗦𝗜𝗔𝗦𝗧𝗦. That inspired me. I've been developing my own SAGE-EXL3 process, and I wanted to see just how far I could push the latest DeepSeek Flash release while keeping the results accessible to the community. So I spent the last 72 hours pushing the limits. And the results genuinely surprised me! 🔥 ━━━━━━━━━━━━━━━━━━━━ 🏆 𝗧𝗛𝗘 𝗥𝗘𝗦𝗨𝗟𝗧𝗦 𝗔𝗥𝗘 𝗖𝗥𝗔𝗭𝗬! I measured my 1.62 bpw expert quant against DeepSeek's native FP4 release checkpoint using held-out sequences and full-vocabulary token predictions. And when I saw the CODE results, I could barely believe it! 🧪 𝗙𝗜𝗗𝗘𝗟𝗜𝗧𝗬 𝗩𝗦 𝗗𝗘𝗘𝗣𝗦𝗘𝗘𝗞'𝗦 𝗡𝗔𝗧𝗜𝗩𝗘 𝗙𝗣𝟴/𝗙𝗣𝟰 𝗥𝗘𝗟𝗘𝗔𝗦𝗘 💻 𝗖𝗢𝗗𝗘 — 𝗛𝘂𝗺𝗮𝗻𝗘𝘃𝗮𝗹 ✅ 99.08% Top-1 Agreement ✅ 𝟵𝟵.𝟲𝟭% 𝗔𝗡𝗦𝗪𝗘𝗥-𝗢𝗡𝗟𝗬 𝗔𝗚𝗥𝗘𝗘𝗠𝗘𝗡𝗧! ✅ KL Divergence: 0.0186 🧠 𝗥𝗘𝗔𝗦𝗢𝗡𝗜𝗡𝗚 — 𝗠𝗔𝗧𝗛-𝟱𝟬𝟬 ✅ 96.04% Top-1 Agreement ✅ 97.98% Answer-Only Agreement ✅ KL Divergence: 0.0349 🔢 𝗠𝗔𝗧𝗛 — 𝗚𝗦𝗠𝟴𝗞 ✅ 95.56% Top-1 Agreement ✅ 94.19% Answer-Only Agreement ✅ KL Divergence: 0.0477 📊 𝗢𝗩𝗘𝗥𝗔𝗟𝗟 𝗛𝗘𝗟𝗗-𝗢𝗨𝗧 𝗙𝗜𝗗𝗘𝗟𝗜𝗧𝗬 🏆 94.52% Top-1 Agreement 📉 Mean KL: 0.0982 🎯 Median KL: 0.00019 𝗡𝗘𝗔𝗥-𝗟𝗢𝗦𝗦𝗟𝗘𝗦𝗦 𝗖𝗢𝗗𝗘 𝗧𝗢𝗞𝗘𝗡 𝗙𝗜𝗗𝗘𝗟𝗜𝗧𝗬 𝗔𝗧 𝗝𝗨𝗦𝗧 𝟭.𝟲𝟮 𝗕𝗣𝗪 𝗘𝗫𝗣𝗘𝗥𝗧𝗦! 🤯 These are direct teacher-forced fidelity measurements against native FP8, not task-level pass rates. But seeing this level of agreement at such aggressive compression is EXACTLY what I was working toward! ━━━━━━━━━━━━━━━━━━━━ ⚡ 𝗔𝗡𝗗 𝗬𝗘𝗦, 𝗜𝗧 𝗔𝗖𝗧𝗨𝗔𝗟𝗟𝗬 𝗥𝗨𝗡𝗦 𝗢𝗡 𝗢𝗡𝗘 𝗦𝗣𝗔𝗥𝗞! Not two Sparks. Not four Sparks. Not an expensive rented GPU cluster. 𝗢𝗡𝗘 𝟭𝟮𝟴𝗚𝗕 𝗗𝗚𝗫 𝗦𝗣𝗔𝗥𝗞. My latest vLLM measurements: ⚡ Decode: 35.5 tok/s 🚀 Maximum measured prefill: 1,189 tok/s For a model this large, compressed this aggressively, and maintaining this level of fidelity... 𝗧𝗛𝗔𝗧'𝗦 𝗔 𝗕𝗜𝗚 𝗗𝗘𝗔𝗟 𝗙𝗢𝗥 𝗟𝗢𝗖𝗔𝗟 𝗔𝗜! 💚 And remember, 1.62 bpw refers to the routed experts, not every parameter in the full model. ━━━━━━━━━━━━━━━━━━━━ 💚 𝗪𝗛𝗬 𝗧𝗛𝗜𝗦 𝗠𝗔𝗧𝗧𝗘𝗥𝗦. I have a simple goal. 𝗠𝗔𝗞𝗘 𝗣𝗢𝗪𝗘𝗥𝗙𝗨𝗟 𝗔𝗜 𝗔𝗖𝗖𝗘𝗦𝗦𝗜𝗕𝗟𝗘 𝗧𝗢 𝗣𝗘𝗢𝗣𝗟𝗘 𝗪𝗛𝗢 𝗗𝗢𝗡'𝗧 𝗛𝗔𝗩𝗘 𝗔𝗖𝗖𝗘𝗦𝗦 𝗧𝗢 𝗗𝗔𝗧𝗔𝗖𝗘𝗡𝗧𝗘𝗥𝗦. I don't want the future of local AI to be determined by who can rent the most GPUs or afford the largest training runs. I want developers, independent researchers and enthusiasts to have the freedom to experiment, quantize, optimize and run powerful models on hardware they actually OWN. NVIDIA's FP4-Local research showed me just how much potential remains in aggressive quantization. My goal with SAGE-EXL3 is to keep pushing that frontier while making advanced quantization more practical for the wider community. Today, I'm sharing the quantized checkpoint and the one-Spark serving recipe. 𝗧𝗛𝗜𝗦 𝗜𝗦 𝗝𝗨𝗦𝗧 𝗧𝗛𝗘 𝗕𝗘𝗚𝗜𝗡𝗡𝗜𝗡𝗚. ━━━━━━━━━━━━━━━━━━━━ 🧪 𝗡𝗢𝗪 𝗙𝗢𝗥 𝗧𝗛𝗘 𝗥𝗘𝗔𝗟 𝗧𝗘𝗦𝗧! The token-level fidelity results are incredible, but I'm not stopping there. I'm currently running a broader evaluation campaign to measure how closely this quant preserves the native FP8 model's actual task performance. 💻 HumanEval+ / MBPP+ 🧠 MMLU-Pro STEM / MATH-500 🔬 SciCode / LiveCodeBench 🔧 BFCL / Multi-step tool chains 🌍 MGSM 𝗧𝗛𝗘 𝗚𝗢𝗔𝗟 𝗜𝗦 𝗧𝗢 𝗣𝗥𝗢𝗩𝗘 𝗧𝗛𝗔𝗧 𝗔𝗚𝗚𝗥𝗘𝗦𝗦𝗜𝗩𝗘 𝗖𝗢𝗠𝗣𝗥𝗘𝗦𝗦𝗜𝗢𝗡 𝗗𝗢𝗘𝗦𝗡'𝗧 𝗛𝗔𝗩𝗘 𝗧𝗢 𝗠𝗘𝗔𝗡 𝗔𝗚𝗚𝗥𝗘𝗦𝗦𝗜𝗩𝗘 𝗤𝗨𝗔𝗟𝗜𝗧𝗬 𝗟𝗢𝗦𝗦! Those task-level results are coming next! 👀 ━━━━━━━━━━━━━━━━━━━━ 👀 𝗔𝗡𝗗 𝗢𝗡𝗘 𝗟𝗜𝗧𝗧𝗟𝗘 𝗧𝗘𝗔𝗦𝗘𝗥... @mr_r0b0t and I have also been working on an GB10 specialized inference engine called 𝗥𝗘𝗫𝗟𝟯. Early testing has been VERY exciting, including 60 tok/s with bursts reaching 74tok/s! We're putting the finishing touches on it and making sure everything is solid before releasing it. More on that VERY soon. 😏 ━━━━━━━━━━━━━━━━━━━━ 🙏 𝗔 𝗛𝗨𝗚𝗘 𝗧𝗛𝗔𝗡𝗞 𝗬𝗢𝗨! To @AWERXAI for supporting my work and helping make this contribution to the community possible. To @nvidiaai for the FP4-Local research that inspired me to push this experiment even further. To @justinn_builds for his initial vllm TP4 recipe that got me started in the right direction for loading this model up. To @mr_r0b0t for collaborating with me on what's coming next! And to @deepseek_ai for continuing to release powerful open-weight models that make community experiments like this possible. 𝗧𝗛𝗘 𝗕𝗘𝗦𝗧 𝗣𝗔𝗥𝗧 𝗢𝗙 𝗢𝗣𝗘𝗡 𝗔𝗜 𝗜𝗦 𝗪𝗛𝗔𝗧 𝗧𝗛𝗘 𝗖𝗢𝗠𝗠𝗨𝗡𝗜𝗧𝗬 𝗖𝗔𝗡 𝗕𝗨𝗜𝗟𝗗 𝗢𝗡 𝗧𝗢𝗣 𝗢𝗙 𝗜𝗧. 💚 🔗 𝗗𝗢𝗪𝗡𝗟𝗢𝗔𝗗 𝗠𝗬 𝗦𝗔𝗚𝗘-𝗘𝗫𝗟𝟯 huggingface.co/vcruz305/DSV4… 🛠️ 𝗢𝗡𝗘-𝗦𝗣𝗔𝗥𝗞 𝗥𝗘𝗖𝗜𝗣𝗘 github.com/vcruz305/DSV4.1-F… 𝗢𝗡𝗘 𝗦𝗣𝗔𝗥𝗞. 𝟭.𝟲𝟮 𝗕𝗣𝗪 𝗘𝗫𝗣𝗘𝗥𝗧𝗦. 𝟵𝟵.𝟲𝟭% 𝗖𝗢𝗗𝗘 𝗔𝗡𝗦𝗪𝗘𝗥 𝗔𝗚𝗥𝗘𝗘𝗠𝗘𝗡𝗧. 𝗟𝗘𝗧'𝗦 𝗞𝗘𝗘𝗣 𝗣𝗨𝗦𝗛𝗜𝗡𝗚 𝗟𝗢𝗖𝗔𝗟 𝗔𝗜 𝗙𝗢𝗥𝗪𝗔𝗥𝗗! 🚀🔥

Post attachment
Cruz avatar
Cruz@ViC305·14h

Hy4 Flash is now released on huggingface!! 🚀🚀🚀 Thanks to @aitrackerbot always being on point! You can finally run Haiku 5.5 locally 😎 Here’s what it had to say about the new model: “Tencent just put the model from Monday's deleted announcement on Hugging Face: • 220B total parameters, but only ~12B active per token, so it's cheap to run • Reads text and images, with a 1M-token context window • In Tencent's blind test with its own users, it scored 2.79 out of 4 vs 2.81 for Claude Haiku 5.5, at the lowest cost per task of any model tested • About 1/20 the cost per task of full Hy4 • Also out: an FP8 version and the pre-trained base model Definitely going to SAGE this one!! Stay tuned 👀 huggingface.co/tencent/Hy4-f…

⚡Meet Hy4 flash, the cheapest model for everyday productivity. Matches Claude Haiku 5.5 (high) on daily work (human blind tests), yet 18x task cost reduction. API pricing (per 1M tokens): $0.003 cached · $0.070 input · $0.281 output. 🔗:hy.tencent.ai/research/hy4-f…

Post attachment
Cruz avatar
Cruz@ViC305·Nov 17

🚨🔥 𝗚𝗣𝗧-𝟲.𝟮 𝗔𝗦𝗧𝗥𝗔 & 𝗦𝗢𝗟 𝗝𝗨𝗦𝗧 𝗥𝗔𝗜𝗦𝗘𝗗 𝗧𝗛𝗘 𝗕𝗔𝗥 𝗔𝗡𝗗 𝗧𝗛𝗘 𝗖𝗬𝗕𝗘𝗥𝗦𝗘𝗖𝗨𝗥𝗜𝗧𝗬 𝗡𝗨𝗠𝗕𝗘𝗥𝗦 𝗔𝗥𝗘 𝗧𝗛𝗘 𝗥𝗘𝗔𝗟 𝗦𝗧𝗢𝗥𝗬! 🧠🛡️ OpenAI is challenging Claude's frontier dominance in professional agents, coding, 3D development AND cybersecurity! 𝗔𝗡𝗗 𝗧𝗛𝗘 𝗡𝗨𝗠𝗕𝗘𝗥𝗦 𝗔𝗥𝗘 𝗪𝗢𝗥𝗧𝗛 𝗟𝗢𝗢𝗞𝗜𝗡𝗚 𝗔𝗧. 👀 ━━━━━━━━━━━━━━━━━━ 🏆 𝗔𝗦𝗧𝗥𝗔 𝘃𝘀 𝗖𝗟𝗔𝗨𝗗𝗘 𝗙𝗔𝗕𝗟𝗘 𝟱.𝟮 🤖 𝗔𝗴𝗲𝗻𝘁𝘀' 𝗟𝗮𝘀𝘁 𝗘𝘅𝗮𝗺 — Professional Agent Work ✅ Astra leads by +5.5 percentage points at just 5% of Fable's cost per task! ⚖️ 𝗛𝗮𝗿𝘃𝗲𝘆 𝗟𝗔𝗕-𝗔𝗔 — Legal Agents ✅ Astra achieves MORE THAN DOUBLE Fable's pass rate at approximately 1/6 the cost! 🎮 𝗔𝗔-𝟯𝗗 — Interactive 3D Development 🥇 Astra: 1,318 Elo • Fable: 1,271 Elo 🔥 +47 Elo for Astra! ━━━━━━━━━━━━━━━━━━ 🧠 𝗔𝗦𝗧𝗥𝗔 𝘃𝘀 𝗖𝗟𝗔𝗨𝗗𝗘 𝗢𝗣𝗨𝗦 𝟱.𝟱 Early Artificial Analysis Intelligence Index figures: 🚀 GPT-6.2 Astra: 62.1 • Claude Opus 5.5: 59.1 Astra leads by 3 points! Claude Fable 5.2 remains ahead at 64.6. ━━━━━━━━━━━━━━━━━━ 🛡️ 𝗕𝗨𝗧 𝗧𝗛𝗘 𝗕𝗜𝗚𝗚𝗘𝗦𝗧 𝗦𝗛𝗜𝗙𝗧 𝗠𝗔𝗬 𝗕𝗘 𝗜𝗡 𝗖𝗬𝗕𝗘𝗥𝗦𝗘𝗖𝗨𝗥𝗜𝗧𝗬. On 𝗘𝘅𝗽𝗹𝗼𝗶𝘁𝗕𝗲𝗻𝗰𝗵 𝗜𝗻𝘁𝗲𝗿𝗻𝗮𝗹 𝗣𝗼𝗿𝘁 — developing working exploits for known V8/Chrome vulnerabilities: 🥇 GPT-6.2 Astra: 53.5% 🥈 GPT-6.2 Sol: 50.5% • GPT-6 Astra: 31.5% 𝗔 𝗠𝗔𝗦𝗦𝗜𝗩𝗘 +𝟮𝟮 𝗣𝗘𝗥𝗖𝗘𝗡𝗧𝗔𝗚𝗘-𝗣𝗢𝗜𝗡𝗧 𝗝𝗨𝗠𝗣 𝗜𝗡 𝗢𝗡𝗘 𝗚𝗘𝗡𝗘𝗥𝗔𝗧𝗜𝗢𝗡! 🤯 Both new models now meet OpenAI's CRITICAL cybersecurity capability threshold. These are internal evaluations conducted with cyber safeguards disabled, not a measure of unrestricted public access. And that's where things get interesting... 🔒 OpenAI is introducing tighter safeguards for higher-risk cyber assistance. 🛡️ GPT-6.2 is planned for OpenAI Daybreak, where separately approved defenders can eventually access supported configurations with reduced cyber refusals for authorized security work. More capability. More restrictions. A growing need for trusted researcher access. ━━━━━━━━━━━━━━━━━━ ⚡ 𝗔𝗡𝗗 𝗚𝗣𝗧-𝟲.𝟮 𝗦𝗢𝗟 𝗜𝗦𝗡'𝗧 𝗝𝗨𝗦𝗧 𝗔 𝗦𝗠𝗔𝗟𝗟𝗘𝗥 𝗔𝗙𝗧𝗘𝗥𝗧𝗛𝗢𝗨𝗚𝗛𝗧. 💻 On FrontierCode 2.0, Sol finishes just 0.5 points behind Astra at roughly ONE-FIFTH the cost per task! API pricing per million input/output tokens: 💰 Astra: $8 / $40 💰 Sol: $2 / $10 ━━━━━━━━━━━━━━━━━━ 𝗠𝗬 𝗕𝗜𝗚𝗚𝗘𝗦𝗧 𝗧𝗔𝗞𝗘𝗔𝗪𝗔𝗬? We're entering a phase where frontier AI isn't just improving at writing code or using tools. It's becoming capable enough that HOW we deploy it, WHO gets access, and HOW efficiently it solves problems are becoming just as important as the leaderboard itself. 𝗧𝗛𝗘 𝗡𝗘𝗫𝗧 𝗙𝗥𝗢𝗡𝗧𝗜𝗘𝗥 𝗜𝗦𝗡'𝗧 𝗝𝗨𝗦𝗧 𝗜𝗡𝗧𝗘𝗟𝗟𝗜𝗚𝗘𝗡𝗖𝗘 — 𝗜𝗧'𝗦 𝗧𝗥𝗨𝗦𝗧𝗘𝗗 𝗖𝗔𝗣𝗔𝗕𝗜𝗟𝗜𝗧𝗬 𝗔𝗧 𝗦𝗖𝗔𝗟𝗘. 🚀🔥

OpenAI@OpenAINov 17

GPT-6.2 Astra is our most capable model ever for cyber. It jumps 22 points over GPT-6 Astra on ExploitBench Internal Port. Due to increased capabilities, we’re launching GPT‑6.2 Astra and Sol with tighter safeguards for higher-risk cybersecurity assistance. We are working to bring these models to Daybreak so approved defenders can access them with reduced refusals.

Post attachment
Cruz avatar
Cruz@ViC305·Nov 16

🚨 ANOTHER GLM FLASH FINE-TUNE HAS HIT HUGGING FACE! 🔥 Meet SAGE S2.1 Flash from @sageai_official & @intern_lm — and these benchmarks are WILD! 🤯 🏆 Beats GLM-5.3-Flash 15/15 🔥 Beats DeepSeek V4.1-Flash 11/15 🚀 Beats Qwen3.8-Max 11/15 Agentic benchmarks are looking INSANE: 💻 MLE-bench Lite (ML coding): 90.3 🖥️ OSWorld (computer use): 85.2 ⌨️ Terminal-Bench 2.1 (terminal/engineering): 87.9 Built on GLM Flash architecture, but punching at frontier-model levels! Naturally, I’ll be creating a SAGE-EXL3 quant of SAGE to fit on ONE DGX Spark! ⚡ SAGE quantizing SAGE? That’s some next-level SAGE wisdom. 😏🧠 🔗 huggingface.co/internlm/Sage…

@sageai_official open releases Sage-S2.1-Flash, a 320B-A18B multimodal MoE that runs and verifies workflows in scientific software. MIT-licensed weights.🤖modelscope.ai/models/Shangha… 🔬 10+ scientific input types. Sage SciPerception unifies molecular graphs, crystal structures, DNA/RNA sequences, spectra, and EEG in a shared representation. 🛠️ Tool use with verification. In the CT demo, it selected 139 slices from 12,487 mixed images, corrected their order after a failed consistency check, and reconstructed the skeleton in 3D Slicer. Other demos cover satellite imagery, CAD assembly checks, and EEG artifact removal. ⚡ ~1/10 the inference cost of Xinghe on the same task, per Shanghai AI Lab. 🏆 Benchmark leads on MLE-bench Lite, OfficeQA Pro, Harvey LAB, and SpreadsheetBench 2 in the lab’s comparison. 🔄 RSI. Task trajectories, tool outputs, and human verification guide the next round of model improvements.

Post attachment
Cruz avatar
Cruz@ViC305·Nov 16

🚨🔥 𝗤𝗪𝗘𝗡𝟰-𝗠𝗔𝗫 𝗜𝗦 𝗢𝗨𝗧 𝗢𝗙 𝗣𝗥𝗘𝗩𝗜𝗘𝗪 𝗡𝗢𝗪 𝗔𝗡𝗗 𝗜𝗧'𝗦 𝗚𝗢𝗜𝗡𝗚 𝗧𝗢𝗘-𝗧𝗢-𝗧𝗢𝗘 𝗪𝗜𝗧𝗛 𝗧𝗛𝗘 𝗙𝗥𝗢𝗡𝗧𝗜𝗘𝗥 𝗚𝗜𝗔𝗡𝗧𝗦! 🤯🚀 The API no longer shows the Preview label, and these benchmark comparisons are seriously exciting! 𝗟𝗢𝗢𝗞 𝗔𝗧 𝗛𝗢𝗪 𝗤𝗪𝗘𝗡𝟰-𝗠𝗔𝗫 𝗦𝗧𝗔𝗖𝗞𝗦 𝗨𝗣! 👀 Across the benchmarks in this chart, Qwen4-Max scores HIGHER than: 🏆 GPT-6.1 Sol — 9/16 benchmarks 🔥 GPT-6 Astra — 7/15 benchmarks ⚡ Claude Opus 5.5 — 6/16 benchmarks 🚀 Kimi K3.1 — 14/16 benchmarks 𝗔𝗡𝗗 𝗦𝗢𝗠𝗘 𝗢𝗙 𝗧𝗛𝗘𝗦𝗘 𝗥𝗘𝗦𝗨𝗟𝗧𝗦 𝗔𝗥𝗘 𝗪𝗜𝗟𝗗! 💻 𝗗𝗘𝗘𝗣𝗦𝗪𝗘 𝘃𝟭.𝟭 — Agentic Coding 🥇 Qwen4-Max: 75.7 • Kimi K3.1: 75.5 • GPT-6.1 Sol: 75.2 • GPT-6 Astra: 74.1 • Claude Opus 5.5: 74.2 𝗤𝗪𝗘𝗡 𝗟𝗘𝗔𝗗𝗦 𝗧𝗛𝗘 𝗘𝗡𝗧𝗜𝗥𝗘 𝗚𝗥𝗢𝗨𝗣! 🧬 𝗕𝗜𝗢𝗠𝗬𝗦𝗧𝗘𝗥𝗬𝗕𝗘𝗡𝗖𝗛 — Scientific Research 🥇 Qwen4-Max: 85.0 • GPT-6.1 Sol: 84.4 • Claude Opus 5.5: 81.9 • Kimi K3.1: 80.6 ⚖️ 𝗛𝗟𝗔𝗕-𝗩𝗔𝗟𝗦 — Legal Agent Work 🥇 Qwen4-Max: 21.3 • Kimi K3.1: 11.7 • Claude Opus 5.5: 9.2 • GPT-6 Astra: 7.1 🤖 𝗔𝗣𝗘𝗫-𝗔𝗚𝗘𝗡𝗧𝗦 — Professional Services • Claude Opus 5.5: 73.5 • Qwen4-Max: 72.1 JUST 1.4 POINTS APART! ⚙️ 𝗔𝗨𝗧𝗢𝗠𝗔𝗧𝗜𝗢𝗡𝗕𝗘𝗡𝗖𝗛 • Kimi K3.1: 73.3 • Qwen4-Max: 72.5 ONLY 0.8 POINTS BEHIND! ━━━━━━━━━━━━━━━━━━ And here's another impressive result: 𝗤𝗪𝗘𝗡𝟰-𝗠𝗔𝗫 𝗢𝗨𝗧𝗦𝗖𝗢𝗥𝗘𝗦 𝗤𝗪𝗘𝗡𝟯.𝟴-𝗠𝗔𝗫 𝗢𝗡 𝗔𝗟𝗟 𝟭𝟱 𝗦𝗛𝗔𝗥𝗘𝗗 𝗕𝗘𝗡𝗖𝗛𝗠𝗔𝗥𝗞𝗦! 🔥 That's a substantial generational improvement. Now, it's not a clean sweep. Terminal-Bench 4.0 and ProgramBench still show meaningful gaps against the competition. But seeing Qwen competing this closely with the biggest frontier models across coding, research, professional agents and automation is exciting! ━━━━━━━━━━━━━━━━━━ 🌍 𝗔𝗡𝗗 𝗛𝗘𝗥𝗘'𝗦 𝗪𝗛𝗬 𝗜 𝗖𝗔𝗥𝗘 𝗦𝗢 𝗠𝗨𝗖𝗛. Qwen4-Max itself is a hosted model, not an open-weight release. But imagine these advances making their way into Qwen's OPEN-WEIGHT lineup! Models we can download, quantize, optimize, and run on our OWN hardware. As someone constantly pushing model compression and local inference, that's what gets me excited. 𝗧𝗛𝗘 𝗙𝗥𝗢𝗡𝗧𝗜𝗘𝗥 𝗜𝗦𝗡'𝗧 𝗝𝗨𝗦𝗧 𝗔𝗕𝗢𝗨𝗧 𝗪𝗛𝗢 𝗛𝗔𝗦 𝗧𝗛𝗘 𝗠𝗢𝗦𝗧 𝗘𝗫𝗣𝗘𝗡𝗦𝗜𝗩𝗘 𝗔𝗣𝗜 𝗔𝗡𝗬𝗠𝗢𝗥𝗘. It's about how much of that intelligence we can eventually bring home. Huge credit to @Alibaba_Qwen and @QwenDevs! 𝗤𝗪𝗘𝗡𝟰 𝗜𝗦 𝗛𝗘𝗥𝗘. 𝗡𝗢𝗪 𝗕𝗥𝗜𝗡𝗚 𝗢𝗡 𝗧𝗛𝗘 𝗢𝗣𝗘𝗡 𝗪𝗘𝗜𝗚𝗛𝗧𝗦! 🚀🔥

📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery! Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window. Highlights: - Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation. - Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables. - Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery. - $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%. The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder. Start building today! 🚀 - Blog: qwen.ai/blog?id=qwen4-max - API: qwencloud.com/models/qwen4-m… - Token Plan: qwencloud.com/pricing/token-… More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀

Post attachment
Cruz avatar
Cruz@ViC305·Nov 15

My one spark Nex-N2.6 quant by @NexEcosystem (a GLM 5.3 Flash finetune) just hit 40tok/s decode and a 1046 tok/s on my stealth inference engine on one DGX Spark 🚀 What do you think is this good for GLM5.3 Flash on ONE spark or should I push it further?! x.com/vic305/status/21216742…

Post attachment
Cruz avatar
Cruz@ViC305·Nov 15

Please do not listen to people who recommend old bitcoin mining chips over DGX Sparks Blackwell chips, it’s ridiculous to say the least. 4x 170HXs will cost you an average of about $140 in the USA and £219 UK in monthly electricity costs alone! VS 2x DGX Sparks about $24 USA and £35 in the UK! Oh and don’t get me started on the cooling that these old bitcoin mining chips need! Let’s not forget it’s 5 year old Ampere tech which lacks native FP4/FP8 Tensor Cores, so it can’t run the low-precision formats that give the Blackwell-based DGX Spark its big inference speed and efficiency edge!! Buy DGX Sparks, buy Blackwell and please do your due diligence before listening to people like this on X it ain’t all about TPS. 🙏🫡

When I first wrote this post, a 170HX was half the price of a DGX Spark. Today, they cost the same. Congratulations if you joined the winning team.

Cruz avatar
Cruz@ViC305·Nov 14

CONGRATULATIONS to the STARSHIP TEAM!! 🚀🎉 They literally CAUGHT a 40 STORY SKYSCAPER BUILDING!! 🤯🤯 Yea you read that right…Starship is as large as a 40 floor skyscraper!! Bravo on making human history! 🇺🇸 THIS IS AMERICA 🇺🇸

Post attachment
SpaceX@SpaceXNov 13

Live webcast of Starship's 15th flight test starts now → x.com/i/broadcasts/2brajgZlO…

Cruz avatar
Cruz@ViC305·Nov 14

🚨🔥 𝗧𝗪𝗢 𝗡𝗘𝗪 𝗦𝗔𝗚𝗘-𝗥𝗘𝗫𝗟𝟯 𝗥𝗘𝗟𝗘𝗔𝗦𝗘𝗦 — 𝗕𝗢𝗧𝗛 𝗔𝗧 𝟮.𝟰𝟴 𝗕𝗣𝗪 𝗙𝗢𝗥 𝗢𝗡𝗘 𝗗𝗚𝗫 𝗦𝗣𝗔𝗥𝗞! 🚀 I ran my updated SAGE quantization process on two 321B-class MoE models, preserving over 95% top-1 agreement with their original FP8 checkpoints! 🧠 𝗡𝗘𝗫-𝗡𝟮.𝟲 — 𝗔𝗚𝗘𝗡𝗧𝗜𝗖 𝗙𝗜𝗡𝗘-𝗧𝗨𝗡𝗘 🏆 95.52% Top-1 agreement 🔬 0.0246 mean KLD 💾 ~96.5 GiB resident Built on GLM-5.3-Flash, Nex-N2.6 focuses on computer use, tool reliability and reducing reasoning loops! Nex-AGI reports improvements over base GLM: ✅ OSWorld: 78.0 → 83.0 ✅ Tool-list adherence: 84.1% → 100% ✅ Reasoning repetition: 6.0% → 1.6% These are upstream model-level results, not separate benchmarks of my quant. ⚡ 𝗚𝗟𝗠-𝟱.𝟯-𝗙𝗟𝗔𝗦𝗛 — 𝗨𝗣𝗗𝗔𝗧𝗘𝗗 𝗦𝗔𝗚𝗘 𝗤𝗨𝗔𝗡𝗧 🏆 95.27% Top-1 agreement 🔬 0.0300 mean KLD 💾 ~96.5 GiB resident 💻 97.03% code token agreement Both are sized for one 128GB DGX Spark, with N2.6 targeting 262K context and its NextN draft resident! 🤗 𝗡𝗘𝗫-𝗡𝟮.𝟲 huggingface.co/vcruz305/Nex-… 🤗 𝗚𝗟𝗠-𝟱.𝟯-𝗙𝗟𝗔𝗦𝗛 huggingface.co/vcruz305/GLM-… Shoutout to @NexEcosystem and @Zai_org for the original models! 𝗧𝗪𝗢 𝗠𝗢𝗗𝗘𝗟𝗦. 𝗢𝗡𝗘 𝗦𝗣𝗔𝗥𝗞. 𝗠𝗢𝗥𝗘 𝗟𝗢𝗖𝗔𝗟 𝗔𝗜 𝗢𝗣𝗧𝗜𝗢𝗡𝗦! 🔥

Post attachment
Cruz avatar
Cruz@ViC305·Nov 12

🚨🌌 𝗤𝗪𝗘𝗡𝟰-𝗠𝗔𝗫-𝗣𝗥𝗘𝗩𝗜𝗘𝗪 𝗕𝗨𝗜𝗟𝗧 𝗠𝗘 𝗔 𝗙𝗨𝗧𝗨𝗥𝗜𝗦𝗧𝗜𝗖 𝟯𝗗 𝗚𝗣𝗨 𝗙𝗟𝗘𝗘𝗧 𝗖𝗢𝗠𝗠𝗔𝗡𝗗 𝗖𝗘𝗡𝗧𝗘𝗥! 🤯🔥 Introducing 𝗖𝗢𝗡𝗦𝗧𝗘𝗟𝗟𝗔𝗧𝗜𝗢𝗡! ✨ Using @NousResearch's Hermes harness, I had Qwen4-Max-Preview build a TRON-inspired command center to visualize and monitor my entire AI compute fleet. ⏱️ 𝟭𝗵 𝟮𝟭𝗺 to first working deployment! 🚀 𝟰𝗵 𝟭𝟯𝗺 total active development! The Hermes run involved: 🧠 430 model turns 🛠️ 525 tool executions 👁️ 33 vision calls 𝗔𝗡𝗗 𝗟𝗢𝗢𝗞 𝗪𝗛𝗔𝗧 𝗜𝗧 𝗕𝗨𝗜𝗟𝗧! 🔥 🖥️ Holographic DGX Sparks, Mac Minis, Mac Studios & GPU servers 🔗 SSH / Tailscale connectivity ⚡ Live inference TPS, decode, prefill & TTFT 📊 GPU usage, VRAM, power, temperature & KV cache 🧠 vLLM, SGLang, EXL3, Ollama, MLX & more 🌐 FastAPI + WebSockets + React + Three.js 📱 Fully responsive desktop & mobile HUD! Real GPU telemetry was tested on a DGX Spark. The public demo uses a clearly labeled simulated fleet; connect your own backend for LIVE monitoring. Huge shoutout to @Alibaba_Qwen, @QwenDevs & @NousResearch! 🌌 𝗧𝗥𝗬 𝗜𝗧: vcruz305.github.io/constella… 💻 𝗢𝗣𝗘𝗡 𝗦𝗢𝗨𝗥𝗖𝗘: github.com/vcruz305/constell… 𝗘𝗩𝗘𝗥𝗬 𝗗𝗘𝗩𝗜𝗖𝗘 𝗔 𝗦𝗧𝗔𝗥. 𝗢𝗡𝗘 𝗖𝗢𝗡𝗦𝗧𝗘𝗟𝗟𝗔𝗧𝗜𝗢𝗡. ✨🚀

1:20
Cruz avatar
Cruz@ViC305·Nov 11

🚨🌍 𝗤𝗪𝗘𝗡𝟰-𝗠𝗔𝗫-𝗣𝗥𝗘𝗩𝗜𝗘𝗪 𝗕𝗨𝗜𝗟𝗧 𝗠𝗘 𝗔 𝗙𝗨𝗟𝗟 𝟯𝗗 𝗘𝗔𝗥𝗧𝗛 𝗠𝗜𝗦𝗦𝗜𝗢𝗡 𝗖𝗢𝗡𝗧𝗥𝗢𝗟 𝗜𝗡 𝗝𝗨𝗦𝗧 𝟭 𝗛𝗢𝗨𝗥 𝗔𝗡𝗗 𝟰𝟳 𝗠𝗜𝗡𝗨𝗧𝗘𝗦! 🤯🚀 Not a mockup. Not a static demo. 𝗔 𝗥𝗘𝗔𝗟, 𝗜𝗡𝗧𝗘𝗥𝗔𝗖𝗧𝗜𝗩𝗘, 𝗣𝗨𝗕𝗟𝗜𝗖𝗟𝗬 𝗗𝗘𝗣𝗟𝗢𝗬𝗘𝗗 𝗔𝗣𝗣𝗟𝗜𝗖𝗔𝗧𝗜𝗢𝗡! Introducing 𝗢𝗥𝗕𝗜𝗧𝗔𝗟. 🌎 I wanted to see just how far @Alibaba_Qwen Qwen4-Max-Preview could push agentic software engineering using @NousResearch's Hermes as my harness. So I gave it an ambitious challenge: build a cinematic Earth intelligence platform combining satellite tracking, rocket launches, aircraft visualization and a fully interactive 3D globe. 𝗙𝗥𝗢𝗠 𝗣𝗥𝗢𝗠𝗣𝗧 𝗧𝗢 𝗟𝗜𝗩𝗘 𝗪𝗘𝗕𝗦𝗜𝗧𝗘: ⏱️ Started: 9:23 AM 🚀 Deployed and verified: 11:10 AM 🔥 Total: 𝟭 𝗛𝗢𝗨𝗥 𝟰𝟳 𝗠𝗜𝗡𝗨𝗧𝗘𝗦! And look at what it built! ━━━━━━━━━━━━━━━━━━ 🌎 𝗖𝗜𝗡𝗘𝗠𝗔𝗧𝗜𝗖 𝟯𝗗 𝗘𝗔𝗥𝗧𝗛 🌍 NASA Blue Marble imagery 🌗 Real Sun-driven day/night terminator 🌃 Nighttime city lights ☁️ Rotating clouds and atmosphere 🌙 Sun, Moon and Milky Way 🎥 Cinematic camera controls 🛰️ 𝗟𝗜𝗩𝗘 𝗦𝗔𝗧𝗘𝗟𝗟𝗜𝗧𝗘 𝗧𝗥𝗔𝗖𝗞𝗜𝗡𝗚 📡 589 objects loaded during testing 🚀 Starlink, GPS, ISS, weather and science 🧮 CelesTrak TLEs + SGP4 propagation 📊 Altitude, velocity and orbital telemetry 🎯 Select the ISS and lock the camera onto it! 🚀 𝗥𝗢𝗖𝗞𝗘𝗧 𝗟𝗔𝗨𝗡𝗖𝗛 𝗠𝗜𝗦𝗦𝗜𝗢𝗡 𝗖𝗢𝗡𝗧𝗥𝗢𝗟 ⏳ 14 upcoming launches loaded during testing 🚀 SpaceX, ULA, Rocket Lab and more 📍 Launchpads, mission data and countdowns 🌍 Click a launch to fly to its launch site! ✈️ 𝗔𝗜𝗥 𝗧𝗥𝗔𝗙𝗙𝗜𝗖 𝗩𝗜𝗦𝗨𝗔𝗟𝗜𝗭𝗔𝗧𝗜𝗢𝗡 320 simulated aircraft on real airport routes when live OpenSky access is blocked by CORS. Clearly labeled as SIMULATED, not falsely presented as live data. ⚡ Time warp up to 300× 🔎 Global search and constellation filters 📱 Desktop and mobile support 🎛️ Full Mission Control HUD ━━━━━━━━━━━━━━━━━━ 💻 𝗧𝗛𝗘 𝗘𝗡𝗚𝗜𝗡𝗘𝗘𝗥𝗜𝗡𝗚 𝗜𝗦 𝗪𝗛𝗔𝗧 𝗥𝗘𝗔𝗟𝗟𝗬 𝗜𝗠𝗣𝗥𝗘𝗦𝗦𝗘𝗗 𝗠𝗘. During that 1h47m session, the agent worked through: ✅ ~1,700 lines of TypeScript/React ✅ CesiumJS 3D Earth rendering ✅ Satellite orbital propagation ✅ Live API integration and CORS handling ✅ Three build/fix cycles ✅ Desktop and mobile browser QA ✅ ISS follow-camera verification ✅ GitHub repository + Pages deployment ✅ Starlink streaming and 2-hour cache hardening And it delivered a working, published application with zero console errors in the reported final browser tests! The stack: React + TypeScript + CesiumJS + Tailwind. ━━━━━━━━━━━━━━━━━━ 🧠 𝗪𝗛𝗔𝗧 𝗔𝗕𝗢𝗨𝗧 𝗧𝗢𝗞𝗘𝗡 𝗨𝗦𝗔𝗚𝗘? The Hermes session ledger recorded approximately: ⚡ 501 API calls 📥 6.07M input tokens ♻️ 51M cache-read tokens 📤 629K output tokens 🧠 381K reasoning tokens within that output IMPORTANT: Those totals cover the entire shared Hermes session, including earlier CubeLab work. ORBITAL's exact token share wasn't separately logged. But ORBITAL itself went from request to deployed application in a measured 107 minutes. ━━━━━━━━━━━━━━━━━━ What excites me isn't that an AI made a pretty interface. It's watching a coding agent coordinate real data sources, 3D graphics, application architecture, debugging, testing and deployment in a single sustained workflow. 𝗙𝗥𝗢𝗠 𝗜𝗗𝗘𝗔 𝗧𝗢 𝗔 𝗪𝗢𝗥𝗞𝗜𝗡𝗚 𝗣𝗥𝗢𝗗𝗨𝗖𝗧 𝗜𝗡 𝗨𝗡𝗗𝗘𝗥 𝗧𝗪𝗢 𝗛𝗢𝗨𝗥𝗦. 🔥 Huge shoutout to @Alibaba_Qwen and @QwenDevs for Qwen4-Max-Preview, and @NousResearch for Hermes! 🌎 𝗧𝗥𝗬 𝗢𝗥𝗕𝗜𝗧𝗔𝗟: vcruz305.github.io/orbital/ 💻 𝗙𝗨𝗟𝗟 𝗦𝗢𝗨𝗥𝗖𝗘: github.com/vcruz305/orbital Find the ISS, lock onto it, accelerate time and explore Earth! 𝗧𝗛𝗜𝗦 𝗜𝗦 𝗧𝗛𝗘 𝗔𝗚𝗘𝗡𝗧𝗜𝗖 𝗘𝗡𝗚𝗜𝗡𝗘𝗘𝗥𝗜𝗡𝗚 𝗙𝗨𝗧𝗨𝗥𝗘 𝗜'𝗩𝗘 𝗕𝗘𝗘𝗡 𝗪𝗔𝗜𝗧𝗜𝗡𝗚 𝗙𝗢𝗥. 🚀🌍

1:04
Cruz avatar
Cruz@ViC305·Nov 11

🔥 EXL3 IS NOT ONE QUANTIZATION RECIPE. QUALITY MATTERS. While the discussion focuses on BF16 embeddings and memory overhead, I want to show the community what I’ve achieved with SAGE, my custom fidelity-first, mixed-K EXL3 quantization process. 🚀 Qwen3.8-27B — SAGE-REXL3 4.50 bpw | 17.3 GB ✅ 96.82% top-1 agreement vs BF16 ✅ 0.0117 KL divergence 3.64 bpw | 14.7 GB ✅ 95.35% top-1 agreement vs BF16 ✅ 0.0321 KL divergence These are full-pack sizes, not numbers that exclude embeddings. Both retain BF16 token embeddings and use a quantized EXL3 K6 lm_head. The difference is how SAGE allocates precision across the model to preserve fidelity. These are measured teacher-forced, full-vocabulary fidelity results against the original BF16 model, not downstream benchmark scores. The format doesn’t determine the quality. The quantization methodology matters. I’d love to see @UnslothAI benchmark both SAGE variants against other formats under identical conditions. Let’s compare real quality at real memory footprints. 🔬 🔗 4.50 bpw: huggingface.co/vcruz305/Qwen… 🔗 3.64 bpw: huggingface.co/vcruz305/Qwen…

Today, we're releasing the highest accuracy Qwen3.8-27B GGUFs. Our new Dynamic V3.1 GGUFs lead by 5–10% on average across Terminal Bench, DeepSWE & more. We’re working on even better version for Qwen4 Run on 12GB RAM. GGUF: huggingface.co/unsloth/Qwen3… Blog: unsloth.ai/docs/models/qwen3…

Post attachment
Cruz avatar
Cruz@ViC305·Nov 11

Woahhh Qwen4-Max Preview is now available on API!! 🚀🚀 I knew we would get something soon!! Going to give this new Max model a go for sure!! 🔥

Qwen4 is coming. Deeper knowledge. Richer world understanding. Stronger coding and agentic capabilities. Built for real-world tasks — from complex coding and knowledge work to front-end builds and professional software workflows. Qwen4-Max is now in preview on Alibaba’s Token Plan. You can also get early access to the new model on QwenWork, Qoder, Qwen APP, and QwenStudio. Put it to the test on a real problem. Show us what you build. 👀 Token Plan: qwencloud.com/pricing/token-…

Post attachment
Cruz avatar
Cruz@ViC305·Nov 10

Ever wanted a simple, quick way to sync your Hermes agents/bots across devices? Introducing Hermes Roam, my @NousResearch Hermes Agent hackathon entry, now in the official Hermes plugin catalog! One push, one pull, and your SOUL, memories, skills and config follow you to every machine. Everything is encrypted and stored on storage you own: an S3/R2 bucket or a private git repo. No service in the middle. Main features: • Secrets stay local: API keys and .env files never leave the machine • No silent overwrites: an out-of-date push is refused until you pull • Every version is kept and restorable • Multi-profile: all your bots travel together • Optional encrypted, read-only archive of your chat sessions • Hands-free auto sync: push when a session ends, pull when the next one starts • Linux, macOS and Windows The video is a real run across my VPS and my Windows laptop 👇 hermes plugins install vcruz305/hermes-roam github.com/vcruz305/hermes-r… hermes-agent.nousresearch.co…

1:29
Cruz avatar
Cruz@ViC305·Nov 9

Fable 5.2 is out and performs just as good on high as Opus 5.5 Max for about 50% less tokens and cost!! 🤯 On xHigh it scored higher than all public models out there…can’t imagine on Max!! Bravo @AnthropicAI !! Now everyone go and sh*t prompt Fable 5.2 to “adversarial review this project and MAKE NO MISTAKES!” 🤣

Fable 5.2 and Mythos 5.2 are a step above Opus 5.5 in almost all the capabilities we evaluated. Testers have found that they communicate more clearly than past Fable and Mythos models. They generate outputs up to 30% faster than Fable 5.1 and Mythos 5.1.

Post attachment