Image created with gemini-3.1-flash-image-preview with claude-sonnet-4-5. Image prompt: Photorealistic 4K landscape of transparent geometric ice sculptures rising from a frozen bay at winter dusk, crystalline structures with visible internal facets connected by ice bridges, sunset gradient from deep blue to warm orange refracting through the sculptures creating prismatic light patterns on surrounding ice and dark water, National Geographic quality nature documentary cinematography, physically grounded with no CGI glow, hyperreal ice textures, contemplative composition with bold sans-serif ‘Open Source’ title text prominently displayed.
❤️ We are partnering with @MiniMax_AI to give Ollama users free usage of MiniMax M2.5 for the next couple of days! ollama run minimax-m2.5:cloud Use MiniMax M2.5 with OpenCode, Claude Code, Codex, OpenClaw via ollama launch! OpenCode: ollama launch opencode –model”” https://x.com/ollama/status/2022018134186791177
Eigent day 0 supports @MiniMax_AI M2.5! Try M2.5 on your open source cowork! With Chinese New Year (Horse) coming, we asked Eigent to generate 10 complete HTML/CSS/JS games (no libraries) across arcade, puzzle, runner, strategy, memory, idle and more. The Developer Agent called”” https://x.com/Eigent_AI/status/2021983494407069926
Introducing M2.5, an open-source frontier model designed for real-world productivity. – SOTA performance at coding (SWE-Bench Verified 80.2%), search (BrowseComp 76.3%), agentic tool-calling (BFCL 76.8%) & office work. – Optimized for efficient execution, 37% faster at complex”” https://x.com/minimax_ai/status/2021980761210134808
MiniMax M2.5 is live now on OpenRouter! @MiniMax_AI’s update to their powerful agentic model M2.1 comes with improved reliability and performance on long running tasks. It’s become a powerful general agent, capable of much more than writing code.”” https://x.com/OpenRouter/status/2021983955898315238
MiniMax M2.5: Built for Real-World Productivity. – MiniMax News | MiniMax https://www.minimax.io/news/minimax-m25
MiniMax’s new open M2.5 and M2.5 Lightning near state-of-the-art while costing 1/20th of Claude Opus 4.6 | VentureBeat https://venturebeat.com/technology/minimaxs-new-open-m2-5-and-m2-5-lightning-near-state-of-the-art-while
MiniMax-M2.5 is a surprising new step in open coding models. The first model where I’ve been able to independently confirm that it’s better than the most recent Claude Sonnet. It showed up in our benchmarks below, and in my vibe checks it felt strong and diverse.”” https://x.com/gneubig/status/2021988250240598108
80.2% on SWE-Bench Verified and 76.3% on BrowseComp is quite impressive. Try @MiniMax_AI M2.5 on @Eigent_AI”” https://x.com/guohao_li/status/2021984827923476922
M2.5 runs at 100 tokens per second. That’s 3x faster than Opus. At $0.06/M blended with caching, you can run subagents in the CLI and just leave them going. Fast models exist. Cheap models exist. Both at SOTA performance is new.”” https://x.com/cline/status/2022034678065373693
US labs are in trouble when it comes to coding If chinese labs can always deliver 90% of the performance for a fifth or a tenth of the price they will capture a significant chunk of the marktet”” https://x.com/scaling01/status/2021636813115535657
GLM-5 was pre-trained on 28.5T tokens and uses DeepSeek Sparse Attention”” https://x.com/scaling01/status/2021627498451370331
@MiniMax_AI M2.5 is now in Cline. + 80.2% SWE-Bench Verified. + 100 tps. $0.06/M blended cost. + 10B activated parameters. And it’s free in Cine for a limited time!”” https://x.com/cline/status/2022034591075512636
🚨Busy week for new models in the Arena: MiniMax M2.5 by @MiniMax_AI is now available in the Text and Code Arena. Bring your toughest prompts and see how it stacks up against the latest models in real-world use. In Battle mode, your votes power the leaderboards. Learn more”” https://x.com/arena/status/2021987555655422257
Honestly I wanna release this beast ASAP — I’m dying to go back to my hometown for Spring Festival 😂 But the more training compute we put in, the more it keeps rising. Painfully happy problem. We hear you guys. M2.5 soon.”” https://x.com/SkylerMiao7/status/2021587213230715306
Instant access to M2.5 on MiniMax Agent web/desktop! @MiniMax_AI”” https://x.com/MiniMaxAgent/status/2021595954143515106
MiniMax M2.5 is now live on BLACKBOX AI. A frontier model designed for real world execution with strong reasoning, reliable tool use, and complex multi step workflows. Engineered for demanding workloads. Ready for production scale orchestration. Switch instantly in the”” https://x.com/blackboxai/status/2022140484601225420
A glance of MiniMax 2.5, are you ready?”” https://x.com/SkylerMiao7/status/2021578926884053084
Congrats @MiniMax_AI! 🎉 Free for 3 days on Qoder, it’s time to put M2.5 through some serious coding sessions!”” https://x.com/qoder_ai_ide/status/2021983111161213365
MiniMax just dropped M2.5 and it’s on par with Opus 4.6 while being 20x cheaper and 3x faster???”” https://x.com/shydev69/status/2021989925143597123
Kimi Agent Swarm blog is here 🐝 https://t.co/XjPeoRVNxG Kimi can spawn a team of specialists to: – Scale output: multi-file generation (Word, Excel, PDFs, slides) – Scale research: parallel analysis of news from 2000-2025 – Scale creativity: a book in 20 writing styles”” https://x.com/Kimi_Moonshot/status/2021141949416362381
Kimi Agent Swarm: 100 Sub-Agents at Scale https://www.kimi.com/blog/agent-swarm
Annoyingly, DeepSeek on the API is still V3 And I can get 128K/2024 claims repeated in previous chats, despite the absence of clues in the preceding context. They’re rolling it out very unevenly, cautiously. I guess it’s a repeat of r1-lite-preview situation. We get a taste.”” https://x.com/teortaxesTex/status/2021515356951695431
DeepSeek finally has frontier level attention. Maybe better than “”frontier””. They announced the plan to solve attention 13 months ago, in V3 paper. They’re making progress.”” https://x.com/teortaxesTex/status/2021578213420405134
DeepSeek has achieved something very Special with attention. I haven’t seen a model that’s so proactive with its context. It doesn’t just have full recall, it *inhabits* a context, feels at home there. It reminds me of Ant hype about Opus self-awareness. Or of test-time training.”” https://x.com/teortaxesTex/status/2021579901548081353
DeepSeek V3.2 & 3.2-Speciale: GPT5-High Open Weights, Context Management, Plans for Compute Scaling | AINews https://news.smol.ai/issues/25-12-01-deepseek-32
Fun to see Deep Think’s real-world impact. Check out how it’s helping researchers catch errors in high-level mathematics research papers. As “”just”” a math undergrad, I couldn’t even dream to do any of this myself!”” https://x.com/OriolVinyalsML/status/2021982723733438725
i think we don’t realize the impact that deepseek had on the open ecosystem, there is so much from them that you can find in almost every frontier open llm today > most of the open frontier models follow the “”finegrain + sparse + shared expert”” deepseek moe recipe > a lot of”” https://x.com/eliebakouch/status/2021577794480644216
i think we don’t realize the impact that deepseek had on the open ecosystem, there is so much from them that you can find in almost every frontier open llm today > most of the open frontier models follow the “”finegrain + sparse + shared expert”” deepseek moe recipe > a lot of”” https://x.com/eliebakouch/status/2021577794480644216?s=46
I’m sorry but if this is DeepSeek-V4 it is unfortunately over”” https://x.com/scaling01/status/2021562929728885166
Tensor Parallelism is killing your DeepSeek-V3 throughput. Period. MLA models only have ONE KV head. If you’re using vanilla TP8, you’re just wasting 7/8 of your VRAM on redundant cache. We just shipped the solution in @sgl_project : 1. DPA (DP Attention): Zero KV redundancy.”” https://x.com/GenAI_is_real/status/2021512872027656344
Within the last few minutes, DeepSeek has been updated. Knowledge cutoff May 2025, context length 1 million tokens. This is likely V4, though it doesn’t admit to being one.”” https://x.com/teortaxesTex/status/2021511733333131311
Short post about Engram, recent paper by DeepSeek: It is essentially very similar to SCONE (link below), where authors train embeddings for a large number of n-grams (e.g. 1B common n-grams like “”Alexander the Great””). [1/2]”” https://x.com/gabriberton/status/2020612533502222459
You can now train MoE models 12× faster with 35% less VRAM via our new Triton kernels (no accuracy loss). Train gpt-oss locally on 12.8GB VRAM. In collab with @HuggingFace, Unsloth trains DeepSeek, Qwen3, GLM faster. Repo: https://t.co/aZWYAtakBP Blog: https://x.com/UnslothAI/status/2021244131927023950
Mistral’s revenue numbers are impressive and growing insanely fast but the bit that’s most exciting as an investor is the durability of this revenue. Mistral selects who they work with carefully to ensure they actually achieve AI transformation.”” https://x.com/paulbz/status/2021537295883481437
Built a helper repo to make MLX Distributed on Apple Silicon way less painful — and used it to run Kimi K-2.5 (658GB on disk) across a 4× Mac Studio cluster over Thunderbolt RDMA. It actually scales. Video is live 👇”” https://x.com/digitalix/status/2021290293715243261
Introducing Kimi K2.5 on Baseten’s Model APIs with the most performant TTFT (0.26 sec) and TPS (340) on Artificial Analysis. Even among a landscape of incredible open source models, Kimi K2.5 stands out with its multi-modal capabilities and it’s ability to accommodate an”” https://x.com/basetenco/status/2021243980802031900
Kimi K2.5 is live on @tinkerapi:”” https://x.com/thinkymachines/status/2020927620872011940
Kimi K2.5 is now live on Qoder. It’s the #1 model on OpenRouter right now, and we’ve got it in AI Chat at 0.3x Credit (Efficient tier). Strong on implementation: SWE-bench Verified 76.8%, great for coding. @Kimi_Moonshot One early user put it well: “”Plan with the Ultimate or”” https://x.com/qoder_ai_ide/status/2020739503812387074
Mooncake originated from a research collaboration between Kimi(Moonshot AI) and Tsinghua University. It was born from the need to solve the ‘memory wall’ in serving massive-scale models like Kimi K-Series. Since open-sourcing, it has evolved into a thriving community-driven”” https://x.com/Kimi_Moonshot/status/2022109533716533612
Kimi K2.5 + Seedance 2 is a perfect workflow. One prompt = A 100MB Excel storyboard generated with images and prompts, which you can use in seedance. I just tested with Hitchcock’s Psycho iconic shower scene. My prompt: I would like to create a derivative work of the”” https://x.com/crystalsssup/status/2021149326290956353
🎙️Nathan Lambert: Open Models Will Never Catch Up https://www.turingpost.com/p/nathanlambert
And so it begins. Looks like / I hope there will be some fresh open-weight models to analyze and write about soon… It’s been a while!”” https://x.com/rasbt/status/2021594284865036637
🤖 From this week’s issue: Official blog post announcing Qwen3-Coder-Next, an 80B-parameter coding model achieving competitive performance on SWE-Bench (70.6% on Verified) while enabling 10x higher throughput for repository-level agentic workflows.”” https://x.com/dl_weekly/status/2021690941879250945
🚀 Introducing Qwen-Image-2.0 — our next-gen image generation model! 🎨 Your imagination, unleashed. ✨ Type a paragraph → get a pro slides ✨ Describe a scene → get photoreal 2K magic ✨ Add text → it just works (no more glitchy letters!) ✨ Key upgrades: ✅ Professional”” https://x.com/Alibaba_Qwen/status/2021137577311600949
A quick update — we’ve fixed a Qwen-Image 2.0 bug in Qwen Chat that impacted: • Classical Chinese poem ordering in image generation • Character consistency during image editing ✅Patch is live now! https://t.co/DWnxVxa0hY Go test it out and drop us your feedback.”” https://x.com/Alibaba_Qwen/status/2021510747671720368
Folks ask about training an RLM like a hypothetical. In the paper, we do post-train and release open-weights RLM-Qwen3-8B-v0.1 on HF. It’s a tiny proof of concept, but it was surprisingly easy to get a marked jump in capability. Maybe learning to recurse is not too hard for 8B.”” https://x.com/lateinteraction/status/2020877152854409691
https://t.co/DIetNHHMp3 Qwen3.5 architecture is out: A vision language model, hybrid SSM-Transformer using Gated DeltaNet linear attention mixed with standard attention, interleaved MRoPE, and shared+routed MoE experts.”” https://x.com/QuixiAI/status/2021109801606893837
Kimi K2‑0905 and Qwen3‑Max preview: two 1T open weights models launched | AINews https://news.smol.ai/issues/25-09-05-1t-models
Qwen https://qwen.ai/blog?id=a6f483777144685d33cd3d2af95136fcbeb57652&from=research.research-list
Qwen https://qwen.ai/blog?id=qwen-image-2.0
Qwen https://qwen.ai/blog?id=qwen-image-layered
Qwen-Image: SOTA text rendering + 4o-imagegen-level Editing Open Weights MMDiT | AINews https://news.smol.ai/issues/25-08-04-qwen-image
An Open Source Dev Kit for AI-native Robotics 📍GitHub: https://t.co/Et5ffb8yW1 Saying hello to me friend and founder of the robot learning company, @JannikGrothusen. Great for beginners! —- Weekly robotics and AI insights. Subscribe free: https://x.com/IlirAliu_/status/2020967022918566114
Open Source Robotic Arm for All Developers [📍Github Below] A robotic arm project (reBot-DevArm) dedicated to lowering the barrier to learning Embodied AI. They focus on “”True Open Source”” (not just the code), they unreservedly open source everything: > Hardware Blueprints:”” https://x.com/IlirAliu_/status/2021661272144584932
Average Throughput of GLM-5 on Openrouter is 14 tps”” https://x.com/scaling01/status/2021981416452764058
Build more. Spend less. GLM-5 is now on YouWare. Landing pages, portfolios, prototypes. All handled fast, with a 200K context window. Save your premium credits for the big builds.”” https://x.com/YouWareAI/status/2021982784948936874
Congrats @Zai_org on GLM-5! Love the permissive MIT license (vs K2.5’s modified MIT). Haven’t chatted with it yet so no vibes, but from the numbers I’m not compelled to switch from @Kimi_Moonshot K2.5: • Similar evals, but GLM-5’s are at bf16 while K2.5’s are at int4 – GLM-5″” https://x.com/QuixiAI/status/2021651135615184988
Day-0 with @Zai_org: GLM-5 is live on DeepInfra 🔥 Built for long-horizon agents that plan, orchestrate, and self-correct. Serving ~100 TPS at launch and as usual the best price on the market!”” https://x.com/DeepInfra/status/2021666854088110318
GLM 5 is 2x the total parameter of GLM 4.5 + deepseek sparse attention for efficient long context this is going to be a crazy model”” https://x.com/eliebakouch/status/2020824645868630065
GLM MoE DSA”” is landing in transformers 👀”” https://x.com/xeophon/status/2020815776890909052
GLM-4.7-Flash-GGUF is now the most downloaded model on @UnslothAI.”” https://x.com/Zai_org/status/2021207517557051627
GLM-5 already available on OpenRouter (with even lower prices)”” https://x.com/scaling01/status/2021637257103651040
GLM-5 has a 200k context length and maximum output of 128k”” https://x.com/scaling01/status/2021628691357298928
GLM-5 is massive. 745B params. LETS FUCKING GOOOOO This should be fun!”” https://x.com/scaling01/status/2020840989947298156
GLM-5 Pricing $1 and $3.2 Output There is also a GLM-5 Code variant that is more expensive👀 almost 8 times cheaper than Opus”” https://x.com/scaling01/status/2021628971939418522
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 @ActuallyIsaak and @kernelpool for the port.”” https://x.com/awnihannun/status/2022007608811696158
https://t.co/ctlyPtiB3j GLM-5 architecture is out: ~740B parameters ~50B active 78 layers, MLA attention lifted from DeepSeek V3, plus DeepSeek V3.2’s sparse attention indexer for 200k context. Basically DeepSeek V3 scale with DSA bolted on.”” https://x.com/QuixiAI/status/2021111352895393960
GLM-5 is out on @huggingface 🔥 > A40B/744B, trained on more tokens (28.5T) > outperforms/on par with closed sota > allows commercial use (MIT licensed) 💗 use with vLLM/SGLang locally or through HF Inference Providers thanks to @novita_labs and @Zai_org 📦”” https://x.com/mervenoyann/status/2021642658188538348
DeepSeek V4-lite, Minimax 2.5, GLM-5 what a bloodbath will Qwen accelerate the release of 3.5?”” https://x.com/teortaxesTex/status/2021586965594857487
Leave a Reply