Learning with Frosty: Lesson #1# I call it DWM-Domain Weight Modification really friendly with non-ablated models assist. Try it out and let me know. ❄️QWEN 2.5 DENSE ARC ONLY - DO NOT USE THIS ON M.O.E.❄️ it will BREAK the model. (Thats lesson #3#)

🚀 FULL GLM-5.3. FOUR DGX SPARKS. A NEW PARETO-FRONTIER CONTENDER. This is a HUGE personal milestone for my SAGE work! 🔥 My GLM-5.3 SAGE EXL3 3.38 bpw brings the full model down to 319 GB while preserving ALL 19,200 routed experts. No pruning. No expert merging. Every bit matters. 🎯 INDEPENDENT-SET FIDELITY • KL divergence: 0.082388 nats • BF16 top-1 agreement: 91.1523% • 10,240 scored positions on text excluded from calibration ⚡ FOUR-SPARK PERFORMANCE Around 41 tok/s average (69 tok/s burst) across six real prompts with DFlash2, with drafted replies matching serial decoding token for token. 256K-context serving demonstrated. Sized toward a 1M-context memory target, with full-length inference validation still ahead. The goal is the highest-fidelity full GLM-5.3 possible on four Sparks WITHOUT sacrificing the memory needed for 256K–1M context. Not just making the weights fit, but leaving room to actually USE the model. 🙏 A MASSIVE thank you to @cfontes, who personally funded the compute required to make this model possible. This release would not have happened without his support! And we’re not done. He’ll be releasing a recipe soon that’s going to blow everyone’s minds! 🤯🔥 This is what community-backed local AI looks like. More capability, less memory, and people coming together to push what’s possible. 💪 Model 👇

RED-SNOW-5.3-FLASH EXL3 is LIVE for both ONE and TWO DGX Sparks. 🥵❄️ ONE Spark: 2.49 bpw, 90.32% top-1, up to 61.7 tok/s TWO Sparks: 4.91 bpw, 96.85% top-1, KL 0.01174 New collab with @Blackfrost_AI. RED-SNOW is an abliteration of GLM-5.3-Flash with a cyber/security fine-tune, built for authorized red-team operations. I took the BF16 release and built two SAGE MixedK EXL3 targets depending on how much hardware you have. 𝗢𝗡𝗘 𝗗𝗚𝗫 𝗦𝗣𝗔𝗥𝗞 2.49 bpw Top-1 agreement vs BF16: 90.32% Mean KLD: 0.1190 Single-Spark decode: ExLlamaV3: 38.7 tok/s TensorFold: 41.0 tok/s REXL3, co-collab stealth engine with @mr_r0b0t : 61.7 tok/s 👀🤫 That is the pack for people trying to squeeze the entire model onto one Spark. 𝗧𝗪𝗢 𝗗𝗚𝗫 𝗦𝗣𝗔𝗥𝗞𝗦 4.91 bpw Top-1 agreement: 96.85% Mean KLD: 0.01174 That is 9,917 / 10,240 positions choosing the same next token as the BF16 reference runtime. This is the quality-first build. SAGE does not just flatten every expert to the same precision. It allocates bits across the model based on sensitivity, which is exactly why I wanted separate 1-Spark and 2-Spark releases instead of forcing one compromise pack on everyone. Both are standalone EXL3 checkpoints. Vision components are packaged. MTP is preserved. Authorized research/security use only. Huge thanks to @Blackfrost_AI for trusting me with another collaboration. And yes, that 61.7 tok/s REXL3 result is something I’ll talk a lot more about soon. 🥶 Each repo has a proper deploy.md file for proper deployment to your Spark setup! Actively working on the 2 spark deploy.md so its validation stage but it will be updated shortly after validation! The one spark deploy.md is already validated! ONE Spark: TWO Sparks: RED-SNOW-5.3-FLASH COLLECTION:

Introducing Beam: a highly efficient agentic open model with 501B total parameters and 23B active. - Frontier reasoning efficiency - Advances the Western open frontier on coding & agentic tasks - Trained end-to-end from scratch Full weights release this month. Learn more about Beam:

🚨 Opus 5.5: Today vs. At Launch — Nerfed? 👀 With everyone talking about a possible nerf, I had to test it myself! I reran the Eiffel Tower benchmark, and honestly, I prefer the launch result. It honestly feels like Opus 5.5 got little nerfed. The tower itself might have slightly better detail now, but the surrounding buildings, trees, river, and overall scene look noticeably worse to me. One run doesn’t prove a nerf, but this comparison surprised me. I want to see whether the same pattern shows up in the night train benchmark next! 🚂🌙 What do you think: nerf or normal variation? 🤔

Deepseek v4.1 vs Qwen-next Same Eiffel Tower prompt, two local models on 2 DGX Sparks, thinking on. DeepSeek V4.1 Flash wrote 114K tokens in 49 min and the code ran first try: real cross-braced lattice, the Seine bend, La Défense in the haze. Qwen3.8 Flash Next was faster (77K tokens in 21 min) but needed 3 code fixes to run, and the river never shows.

🚨 Opus 5.5: Today vs. At Launch — Nerfed? 👀 With everyone talking about a possible nerf, I had to test it myself! I reran the Eiffel Tower benchmark, and honestly, I prefer the launch result. It honestly feels like Opus 5.5 got little nerfed. The tower itself might have slightly better detail now, but the surrounding buildings, trees, river, and overall scene look noticeably worse to me. One run doesn’t prove a nerf, but this comparison surprised me. I want to see whether the same pattern shows up in the night train benchmark next! 🚂🌙 What do you think: nerf or normal variation? 🤔