Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

Alibaba just open-sourced Qwen 3.6-35B A3B 35B MoE model (only ~3B active), multimodal, agentic coding, Apache 2.0. This isn’t just efficient—it’s built for real workflows. Open models are catching up fast.

Mervin Praison

11,380 subscribers

13,063 просмотров • 5 месяцев назад •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

Introducing 0GM-1.0-35B-A3B. Our first proprietary AI model. Mixture of Experts (MoE), 35B parameters, 3B active per token. Trained on our own decentralized GPU network. Open source under Apache 2.0.

Introducing 0GM-1.0-35B-A3B. Our first proprietary AI model. Mixture of Experts (MoE), 35B parameters, 3B active per token. Trained on our own decentralized GPU network. Open source under Apache 2.0.

0G Labs (Home of Infinite AI)

48,838 просмотров • 5 месяцев назад

Qwen3.8-Max is not just another AI model. Alibaba’s flagship model packs: → 2.4T parameters → 1M-token context window → Native vision + text → Sparse MoE with only 95B active parameters But the real flex? It can handle autonomous coding, long-horizon tasks, massive documents, research workflows, and multimodal reasoning. And it’s now available globally through Alibaba Cloud Model Studio API. Qwen is clearly going after the agentic AI era. And it’s now available globally through Alibaba Cloud Model Studio. Alibaba Group Qwen Qwen Studio: API: Blog: Alizila: Website link :

Qwen3.8-Max is not just another AI model. Alibaba’s flagship model packs: → 2.4T parameters → 1M-token context window → Native vision + text → Sparse MoE with only 95B active parameters But the real flex? It can handle autonomous coding, long-horizon tasks, massive documents, research workflows, and multimodal reasoning. And it’s now available globally through Alibaba Cloud Model Studio API. Qwen is clearly going after the agentic AI era. And it’s now available globally through Alibaba Cloud Model Studio. Alibaba Group Qwen Qwen Studio: API: Blog: Alizila: Website link :

Jawad Rahman

64,780 просмотров • 2 месяцев назад

Fully Aggressive variant Uncensored MoE model that won’t refuse red team or pentesting tasks & coding. - 35B parameters, 3B active (MoE) - 0 refusals (Aggressive variant) - Excellent agentic coding + multimodal Powerful uncensored coding model 73.4% on SWE-bench Verified & fully uncensored, It dangerous (in a good way). This is currently one of the best options if you want power without guardrails. High quality imatrix GGUF quants available

Fully Aggressive variant Uncensored MoE model that won’t refuse red team or pentesting tasks & coding. - 35B parameters, 3B active (MoE) - 0 refusals (Aggressive variant) - Excellent agentic coding + multimodal Powerful uncensored coding model 73.4% on SWE-bench Verified & fully uncensored, It dangerous (in a good way). This is currently one of the best options if you want power without guardrails. High quality imatrix GGUF quants available

Md Ismail Šojal 🕷️

28,103 просмотров • 3 месяцев назад

A major flagship addition to the open-source AI Models. Tencent has officially launched Hunyuan Hy3. This could be the most interesting open-source AI releases this year. ✓ 295B MoE architecture ✓ 21B active parameters ✓ 256K context ✓ Strong coding & reasoning ✓ Built for agent workflows ✓ Apache 2.0 license A big step forward for open AI infrastructure. Tencent Hy #Hy3 #Hunyuan #OpenSourceAI

A major flagship addition to the open-source AI Models. Tencent has officially launched Hunyuan Hy3. This could be the most interesting open-source AI releases this year. ✓ 295B MoE architecture ✓ 21B active parameters ✓ 256K context ✓ Strong coding & reasoning ✓ Built for agent workflows ✓ Apache 2.0 license A big step forward for open AI infrastructure. Tencent Hy #Hy3 #Hunyuan #OpenSourceAI

ZOYA ✪

45,196 просмотров • 3 месяцев назад

Qwen just quietly dropped the most dangerous open model of 2026 Qwen3.6-Plus. native 1M context. agentic workflows. multimodal. half the size of competitors. and it's free on OpenRouter right now. you're not ready for this breakdown 🧵

Qwen just quietly dropped the most dangerous open model of 2026 Qwen3.6-Plus. native 1M context. agentic workflows. multimodal. half the size of competitors. and it's free on OpenRouter right now. you're not ready for this breakdown 🧵

Farhan

53,820 просмотров • 6 месяцев назад

Meet Nex-N2.5 family, our latest open source agentic models. Mini (35B) and Pro (397B) bring multimodal understanding and computer use. Max (1.6T) is a text-only MoE model built for complex reasoning, coding, and agent workflows. Strong benchmark results in workflow automation and computer use: ⚡ Max scores 50.2 on AutomationBench v1.0.6—just 0.1 points behind Claude Opus 5 ⚡ Pro scores 56.4 on OSWorld-2, ahead of Qwen3.8-Max at 46.7 With continuous action, visual feedback, and self-correction, NEX-N2.5 can work across Blender and CAD tools, handle expense reimbursements, and even play PC games—turning visual understanding into sustained, real-world execution. 🔗Hugging Face 🔗@modelscope 🔗 Github 🔗Website

Meet Nex-N2.5 family, our latest open source agentic models. Mini (35B) and Pro (397B) bring multimodal understanding and computer use. Max (1.6T) is a text-only MoE model built for complex reasoning, coding, and agent workflows. Strong benchmark results in workflow automation and computer use: ⚡ Max scores 50.2 on AutomationBench v1.0.6—just 0.1 points behind Claude Opus 5 ⚡ Pro scores 56.4 on OSWorld-2, ahead of Qwen3.8-Max at 46.7 With continuous action, visual feedback, and self-correction, NEX-N2.5 can work across Blender and CAD tools, handle expense reimbursements, and even play PC games—turning visual understanding into sustained, real-world execution. 🔗Hugging Face 🔗@modelscope 🔗 Github 🔗Website

Nex

261,477 просмотров • 1 месяц назад

Qwen 3.6 Plus Preview just dropped on OpenRouter for Free. > 1M token context. > $0 input. $0 output. Meanwhile: - Claude Opus 4.6 → $5 / $25 per 1M - GPT-5.4 → charges for 1M context Qwen just nuked the pricing model. Open source isn’t slowing down. It’s taking over.

Qwen 3.6 Plus Preview just dropped on OpenRouter for Free. > 1M token context. > $0 input. $0 output. Meanwhile: - Claude Opus 4.6 → $5 / $25 per 1M - GPT-5.4 → charges for 1M context Qwen just nuked the pricing model. Open source isn’t slowing down. It’s taking over.

WorldofAI

146,277 просмотров • 6 месяцев назад

MiniMax M2.1 is officially live🚀 Built for real-world coding and AI-native organizations — from vibe builds to serious workflows. A SOTA 10B-activated OSS coding & agent model, scoring 72.5% on SWE-multilingual and 88.6% on our newly open-sourced VIBE-bench, exceeding leading closed-source models like Gemini 3 Pro and Claude 4.5 Sonnet. The most powerful OSS model for the agentic era is here.

MiniMax M2.1 is officially live🚀 Built for real-world coding and AI-native organizations — from vibe builds to serious workflows. A SOTA 10B-activated OSS coding & agent model, scoring 72.5% on SWE-multilingual and 88.6% on our newly open-sourced VIBE-bench, exceeding leading closed-source models like Gemini 3 Pro and Claude 4.5 Sonnet. The most powerful OSS model for the agentic era is here.

MiniMax (official)

1,074,643 просмотров • 9 месяцев назад

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools

Toha Khan

123,375 просмотров • 7 месяцев назад

We just released Gemma 4 — our most intelligent open models to date. Built from the same world-class research as Gemini 3, Gemma 4 brings breakthrough intelligence directly to your own hardware for advanced reasoning and agentic workflows. Released under a commercially permissive Apache 2.0 license so anyone can build powerful AI tools. 🧵↓

We just released Gemma 4 — our most intelligent open models to date. Built from the same world-class research as Gemini 3, Gemma 4 brings breakthrough intelligence directly to your own hardware for advanced reasoning and agentic workflows. Released under a commercially permissive Apache 2.0 license so anyone can build powerful AI tools. 🧵↓

Google

7,760,271 просмотров • 6 месяцев назад

Qwen3.8-Max is built for more than just answering prompts. Alibaba’s newest flagship model brings: • 2.4T total parameters • 1M-token context • Native vision + text reasoning • 95B active parameters through Sparse MoE The interesting part is what it can actually do. From autonomous coding and long-running tasks to analyzing 200+ page documents and massive streams of information, Qwen3.8-Max is designed to work through complex tasks, not just generate text. Now accessible globally through Alibaba Cloud Model Studio. Alibaba Group Qwen Qwen Studio: API: Blog: Alizila:

Qwen3.8-Max is built for more than just answering prompts. Alibaba’s newest flagship model brings: • 2.4T total parameters • 1M-token context • Native vision + text reasoning • 95B active parameters through Sparse MoE The interesting part is what it can actually do. From autonomous coding and long-running tasks to analyzing 200+ page documents and massive streams of information, Qwen3.8-Max is designed to work through complex tasks, not just generate text. Now accessible globally through Alibaba Cloud Model Studio. Alibaba Group Qwen Qwen Studio: API: Blog: Alizila:

Samaira AI

81,746 просмотров • 2 месяцев назад

$Someone is running Qwen3.8-35B-A3B on a single RTX 3060 with just 12GB of VRAM and 16GB of system RAM. And the reported numbers are surprisingly good. ~50 tok/s decode ~550 tok/s prefill 262K context window 12GB VRAM 16GB system RAM No multi-GPU setup. Just a consumer graphics card and a carefully configured local inference engine. The model is particularly interesting because it’s Qwen3.8 distilled into the Qwen3.6-35B-A3B architecture. That makes it worth investigating beyond raw inference speed. The developer currently uses Ornith-1.5 as the engine behind their Hermes and Pi agents. Now the question is whether this new model can deliver enough quality and reliability to replace it. The next step is testing it as an agent, not just a chatbot. The planned comparison includes four models: • Qwen3.8-35B-A3B • Ornith-1.5-35B • Qwen3.8-27B • Base Qwen3.6-35B-A3B An agentic coding benchmark should help reveal whether the distilled model can follow instructions, modify code, solve problems, and complete multi-step tasks effectively. That distinction matters because high token throughput doesn’t automatically translate into a better coding agent. A model can generate code quickly and still struggle with planning, debugging, tool calls, or completing a task without intervention. The hardware configuration is also worth examining. The developer is using llama-server with the Q4_K_M GGUF quantization, Flash Attention enabled, and 26 MoE layers assigned to CPU processing. The configuration also enables a 262,144-token context window, Q8 KV-cache quantization, and reasoning mode. This is how the setup is configured: llama-server \ -m Qwen3.8-35B-A3B-Q4_K_M.gguf \ -ngl 99 \ --n-cpu-moe 26 \ -c 262144 \ -fa on \ --jinja \ -np 1 \ --cache-type-k q8_0 \ --cache-type-v q8_0 \ --temp 0.6 \ --top-p 0.95 \ --top-k 20 \ --reasoning on One important detail: the model’s 35B parameter count doesn’t mean all 35 billion parameters are activated for every token. The A3B designation indicates roughly 3 billion active parameters per token. That sparse MoE architecture helps explain how a model of this size can deliver high generation speeds on modest hardware. CPU-based expert processing also helps work around the GPU’s limited memory, although the exact trade-offs depend on the workload. The 262K context claim is equally interesting. Supporting a context window that large doesn’t mean every request at that length will maintain the same speed. Prompt length, KV-cache memory, and CPU/GPU placement can all affect performance. The developer’s next test is particularly useful: graphing decode speed against context length. That should help show how performance changes as the prompt grows, rather than presenting one speed figure without the conditions behind it. If the coding benchmark also holds up, this could become a serious option for people who want capable local agents without investing in expensive GPUs. For now, the reported inference numbers are promising. The real test is whether Qwen3.8-35B-A3B can match or outperform Ornith-1.5-35B on actual agentic coding tasks. Fast local inference gets your attention. Reliable autonomous task completion is what earns a model a place in the stack.$

Someone is running Qwen3.8-35B-A3B on a single RTX 3060 with just 12GB of VRAM and 16GB of system RAM. And the reported numbers are surprisingly good. ~50 tok/s decode ~550 tok/s prefill 262K context window 12GB VRAM 16GB system RAM No multi-GPU setup. Just a consumer graphics card and a carefully configured local inference engine. The model is particularly interesting because it’s Qwen3.8 distilled into the Qwen3.6-35B-A3B architecture. That makes it worth investigating beyond raw inference speed. The developer currently uses Ornith-1.5 as the engine behind their Hermes and Pi agents. Now the question is whether this new model can deliver enough quality and reliability to replace it. The next step is testing it as an agent, not just a chatbot. The planned comparison includes four models: • Qwen3.8-35B-A3B • Ornith-1.5-35B • Qwen3.8-27B • Base Qwen3.6-35B-A3B An agentic coding benchmark should help reveal whether the distilled model can follow instructions, modify code, solve problems, and complete multi-step tasks effectively. That distinction matters because high token throughput doesn’t automatically translate into a better coding agent. A model can generate code quickly and still struggle with planning, debugging, tool calls, or completing a task without intervention. The hardware configuration is also worth examining. The developer is using llama-server with the Q4_K_M GGUF quantization, Flash Attention enabled, and 26 MoE layers assigned to CPU processing. The configuration also enables a 262,144-token context window, Q8 KV-cache quantization, and reasoning mode. This is how the setup is configured: llama-server \ -m Qwen3.8-35B-A3B-Q4_K_M.gguf \ -ngl 99 \ --n-cpu-moe 26 \ -c 262144 \ -fa on \ --jinja \ -np 1 \ --cache-type-k q8_0 \ --cache-type-v q8_0 \ --temp 0.6 \ --top-p 0.95 \ --top-k 20 \ --reasoning on One important detail: the model’s 35B parameter count doesn’t mean all 35 billion parameters are activated for every token. The A3B designation indicates roughly 3 billion active parameters per token. That sparse MoE architecture helps explain how a model of this size can deliver high generation speeds on modest hardware. CPU-based expert processing also helps work around the GPU’s limited memory, although the exact trade-offs depend on the workload. The 262K context claim is equally interesting. Supporting a context window that large doesn’t mean every request at that length will maintain the same speed. Prompt length, KV-cache memory, and CPU/GPU placement can all affect performance. The developer’s next test is particularly useful: graphing decode speed against context length. That should help show how performance changes as the prompt grows, rather than presenting one speed figure without the conditions behind it. If the coding benchmark also holds up, this could become a serious option for people who want capable local agents without investing in expensive GPUs. For now, the reported inference numbers are promising. The real test is whether Qwen3.8-35B-A3B can match or outperform Ornith-1.5-35B on actual agentic coding tasks. Fast local inference gets your attention. Reliable autonomous task completion is what earns a model a place in the stack.

FHILY👑

19,254 просмотров • 2 дней назад

Alibaba just dropped Qwen3.5-397B-A17B and there's a lot to unpack. 397B params, 17B active per forward pass. Sparse MoE done right. But the real story isn't the size—it's the architecture choices. The MoE Design Most MoE models feel like bolt-ons. Qwen 3.5's sparse activation is native—only 4.3% of parameters fire per token. That's how you get trillion-parameter-class performance without trillion-parameter inference costs. The 0.8 RMB/million tokens pricing isn't subsidized; it's structurally earned. Native Multimodal, Not Glued-On This is a vision-language model from the ground up. Heterogeneous architecture—separate processing pipelines for text, image, video that fuse early. Not a vision encoder slapped onto an LLM. The result: 90.8 on OmniDocBench, 79.0 on MMMU-Pro. Document understanding and visual reasoning without the usual brittleness. The Context Window Reality Qwen3.5-Plus (the hosted version) ships with 1M tokens by default. That's not a marketing number—they're actually positioning it for long-document workflows. With built-in adaptive tool use, it's clearly aimed at agentic automation, not just chat. What Actually Impressed Me • FP8 native pipeline: ~50% activation memory reduction • Async RL framework for continuous refinement—training and inference workloads separated • 201 languages (up from 119), 250k vocab for better low-resource encoding • Apache 2.0 license. Full weights on HuggingFace and ModelScope. The Benchmark Context 76.4 on SWE-bench Verified puts it in the range where it can handle real debugging workflows. 72.9 on BFCL v4 for agentic tool use. 88.4 on GPQA Diamond. These aren't SOTA in isolation, but the breadth is unusual—strong across reasoning, coding, multimodal, and agentic tasks. The Honest Caveat I haven't stress-tested the 1M context for needle-in-haystack retrieval yet. And "native multimodal" claims need real-world torture testing—PDFs with tables, charts, mixed layouts. Benchmarks are benchmarks. Bottom Line This isn't just another model release. It's a bet on efficient scale: big model capabilities, small active compute, open weights. At 1/18th the cost of Gemini 3 Pro, it's going to force pricing conversations across the board.

Alibaba just dropped Qwen3.5-397B-A17B and there's a lot to unpack. 397B params, 17B active per forward pass. Sparse MoE done right. But the real story isn't the size—it's the architecture choices. The MoE Design Most MoE models feel like bolt-ons. Qwen 3.5's sparse activation is native—only 4.3% of parameters fire per token. That's how you get trillion-parameter-class performance without trillion-parameter inference costs. The 0.8 RMB/million tokens pricing isn't subsidized; it's structurally earned. Native Multimodal, Not Glued-On This is a vision-language model from the ground up. Heterogeneous architecture—separate processing pipelines for text, image, video that fuse early. Not a vision encoder slapped onto an LLM. The result: 90.8 on OmniDocBench, 79.0 on MMMU-Pro. Document understanding and visual reasoning without the usual brittleness. The Context Window Reality Qwen3.5-Plus (the hosted version) ships with 1M tokens by default. That's not a marketing number—they're actually positioning it for long-document workflows. With built-in adaptive tool use, it's clearly aimed at agentic automation, not just chat. What Actually Impressed Me • FP8 native pipeline: ~50% activation memory reduction • Async RL framework for continuous refinement—training and inference workloads separated • 201 languages (up from 119), 250k vocab for better low-resource encoding • Apache 2.0 license. Full weights on HuggingFace and ModelScope. The Benchmark Context 76.4 on SWE-bench Verified puts it in the range where it can handle real debugging workflows. 72.9 on BFCL v4 for agentic tool use. 88.4 on GPQA Diamond. These aren't SOTA in isolation, but the breadth is unusual—strong across reasoning, coding, multimodal, and agentic tasks. The Honest Caveat I haven't stress-tested the 1M context for needle-in-haystack retrieval yet. And "native multimodal" claims need real-world torture testing—PDFs with tables, charts, mixed layouts. Benchmarks are benchmarks. Bottom Line This isn't just another model release. It's a bet on efficient scale: big model capabilities, small active compute, open weights. At 1/18th the cost of Gemini 3 Pro, it's going to force pricing conversations across the board.

Bo Wang

13,221 просмотров • 7 месяцев назад

NVIDIA Releases Audex (Nemotron-Labs-Audex-30B-A3B): A Unified Audio-Text LLM That Preserves the Text Intelligence of Its Backbone Most unified audio models pay a text tax. Add audio output, and reasoning benchmarks drop — even when the only new output is speech. NVIDIA just released one that doesn't. Audex (Nemotron-Labs-Audex-30B-A3B) is a 30B MoE with 3B active parameters, built on the text-only Nemotron-Cascade-2-30B-A3B backbone. Audio inputs are projected into the text embedding space. Text tokens and quantized audio tokens are then generated the same way, inside one MoE decoder — no thinker–talker split, no stacked cascade. Here's what's actually interesting: → One model, audio in and out: understanding, ASR, translation, TTS, text-to-audio, speech-to-speech → Text holds vs its own backbone: IMO AnswerBench 81.1 vs 79.3, MMLU-Redux 86.4 vs 86.3 → Beats text-only Qwen3.5-35B-A3B on several tasks: LiveCodeBench v6 85.3 vs 74.6, IFBench 77.8 vs 70.2 → The usual tax, for contrast: Qwen3-Omni-30B-A3B-Thinking drops to 60.4 on HMMT vs 71.4 for its text backbone → 6.82 WER on OpenASR, ahead of Step-Audio-R1.1-33B (7.91) and Qwen3-Omni-Thinking (8.00) → Two codecs: X-Codec2 for speech (50 tok/s, FSQ, 65,536 codebook), X-Codec for general audio (200 tok/s, 4 flattened RVQ layers) — the only strong open model generating general audio beyond speech Full analysis: Paper: Model weights: NVIDIA AI NVIDIA Wei Ping

NVIDIA Releases Audex (Nemotron-Labs-Audex-30B-A3B): A Unified Audio-Text LLM That Preserves the Text Intelligence of Its Backbone Most unified audio models pay a text tax. Add audio output, and reasoning benchmarks drop — even when the only new output is speech. NVIDIA just released one that doesn't. Audex (Nemotron-Labs-Audex-30B-A3B) is a 30B MoE with 3B active parameters, built on the text-only Nemotron-Cascade-2-30B-A3B backbone. Audio inputs are projected into the text embedding space. Text tokens and quantized audio tokens are then generated the same way, inside one MoE decoder — no thinker–talker split, no stacked cascade. Here's what's actually interesting: → One model, audio in and out: understanding, ASR, translation, TTS, text-to-audio, speech-to-speech → Text holds vs its own backbone: IMO AnswerBench 81.1 vs 79.3, MMLU-Redux 86.4 vs 86.3 → Beats text-only Qwen3.5-35B-A3B on several tasks: LiveCodeBench v6 85.3 vs 74.6, IFBench 77.8 vs 70.2 → The usual tax, for contrast: Qwen3-Omni-30B-A3B-Thinking drops to 60.4 on HMMT vs 71.4 for its text backbone → 6.82 WER on OpenASR, ahead of Step-Audio-R1.1-33B (7.91) and Qwen3-Omni-Thinking (8.00) → Two codecs: X-Codec2 for speech (50 tok/s, FSQ, 65,536 codebook), X-Codec for general audio (200 tok/s, 4 flattened RVQ layers) — the only strong open model generating general audio beyond speech Full analysis: Paper: Model weights: NVIDIA AI NVIDIA Wei Ping

Marktechpost AI

28,301 просмотров • 3 месяцев назад

Parameter scaling just broke in front of me. I gave 90 math problems as images to 10 local vision models, 3 runs each, only consistent answers across all 3 runs counted. Two findings: > Gemma 4 was the most consistent family, 31B took the crown at 89.6%. > But Qwen 3.5 4B came 2 answers behind it. A 4B. Ranked 2nd out of 10. Vision-math isn't one skill, it's two: reading the image, then solving it. The real lesson for anyone running local: small ≠ weak. If you're building agentic workflows, finding the right model for each task matters more than reaching for the biggest one. In this test, the 4B ran far faster because of its size, scored higher, and left VRAM free for the rest of your stack. Full results: 🥇 Gemma 4 31B — 242/270 (89.6%) 🥈 Qwen 3.5 4B — 240/270 (88.9%) 🥉 Gemma 4 E4B — 222/270 (82.2%) 🥉 Qwen 3.6 27B — 222/270 (82.2%) 5. Gemma 4 26B A4B — 216/270 (80.0%) 6. Qwen 3.5 2B — 201/270 (74.4%) 7. Gemma 4 E2B — 192/270 (71.1%) 8. Qwen 3.6 35B A3B — 192/270 (71.1%) 9. Qwen 3.5 9B — 168/270 (62.2%) 10. Qwen 3.5 0.8B — 45/270 (16.7%) All GGUF + mmproj, Unsloth's Q6_K_XL quant.

Parameter scaling just broke in front of me. I gave 90 math problems as images to 10 local vision models, 3 runs each, only consistent answers across all 3 runs counted. Two findings: > Gemma 4 was the most consistent family, 31B took the crown at 89.6%. > But Qwen 3.5 4B came 2 answers behind it. A 4B. Ranked 2nd out of 10. Vision-math isn't one skill, it's two: reading the image, then solving it. The real lesson for anyone running local: small ≠ weak. If you're building agentic workflows, finding the right model for each task matters more than reaching for the biggest one. In this test, the 4B ran far faster because of its size, scored higher, and left VRAM free for the rest of your stack. Full results: 🥇 Gemma 4 31B — 242/270 (89.6%) 🥈 Qwen 3.5 4B — 240/270 (88.9%) 🥉 Gemma 4 E4B — 222/270 (82.2%) 🥉 Qwen 3.6 27B — 222/270 (82.2%) 5. Gemma 4 26B A4B — 216/270 (80.0%) 6. Qwen 3.5 2B — 201/270 (74.4%) 7. Gemma 4 E2B — 192/270 (71.1%) 8. Qwen 3.6 35B A3B — 192/270 (71.1%) 9. Qwen 3.5 9B — 168/270 (62.2%) 10. Qwen 3.5 0.8B — 45/270 (16.7%) All GGUF + mmproj, Unsloth's Q6_K_XL quant.

stevibe

22,882 просмотров • 4 месяцев назад

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.

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 просмотров • 3 месяцев назад