Pinned Tweet
ToolBraid Companion is live on the Microsoft Store 💪🏻
apps.microsoft.com/detail/9P…
Built for developers. Browser tools + WebMCP on supported sites, straight from your Codex session.
Free. You choose which sites your AI can access.
Who’s up for trying it? 👀
3
371
Unreleased beta engine at 128K context.
A single RTX 5090. 🚀
I've pushed Qwen 27B NVFP4 on #TensorFold to previous 309 to 326.93 tokens/ in a real cold-decode test.
Part that matters most: 2,048 generated tokens.
Byte-for-byte identical output. Zero output drift.
~344 tokens/sec average decode. ONE RTX 5090.
26 minutes of local AI coding in Claude Code.
THREE automatic context compactions.
Qwen3.8 27b NVFP4 - The agent kept working. 🔥
Experimental Unreleased BLACKFURY Exact on #TensorFold
Setup + independent test results below 🧵
2/5
The task: build BenchLedger, an offline benchmark registry that syncs data across machines.
SQLite migration. CSV/JSON imports. Duplicate events. Replica conflicts. Crash recovery. Reports.
The challenge: keeping the data consistent through all of it.
2
52
@SandhuJonathan does this real agentic work satisfying your expectations?
1
20
4/5
Across the run:
• 169 successful requests
• 277,887 generated tokens
• ~2,574 tok/s fresh-token prefill
The ~344 tok/s decode average is total output divided by generation time. It excludes prefill and tool execution. Context length varied.
1
30
BLACKFURY Exact V1.0.1 Released - built on #TensorFold by @ashxhart 🚀
Qwen3.8-27B NVFP4 on one RTX 5090: 309.53 tok/s decode at 131K context in a single stream
+10.20% vs the previous release, comparing medians from separate runs.
Code 👇
Qwen3.8-27B NVFP4 + DFlash2 running locally on a single RTX 5090 - powered by #TensorFold 🔥
⚡ 753.3 tok/s peak live
📊 376.5 tok/s average decode across 36 requests 🧠 Thinking LOW
🛠️ Real coding, browser checks and visual fixes
Peak at 5:21 - even Goku powers up 😂
100
Qwen 4 27b it's around the corner 🔥
Can't wait to play with it.
Exciting news for #LocalAI
📢 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. 👀
1
91
734 tok/s coding peak on ONE RTX 5090 🔥
330 tok/s average decode in the same loop.
BLACKFURY Exact: my #TensorFold variant optimized for single stream.
Qwen3.8-27B NVFP4 + DFlash2. 128K window.
Peak sampled over ~1s.
Built on @ashxhart's work. Code + results below 🧵
Parallel still works: 12 synchronized branches, thinking OFF.
790.23 tok/s across the generation batch.
1,079.12 tok/s sampled aggregate peak.
Different workload and context from the single-stream test. These are combined rates across all 12 branches.
1
175
Source, patches, setup and measurements:
github.com/Maharajahu/qwen-p…
One 32 GB GPU, with active inference entirely in VRAM.
If you try it on your 5090, I'd love to see your results. #LocalAI
3
216
RTX 5090. 12 branches. Over 1,000 tok/s🤯
Qwen3.8-27B NVFP4 locally on my experimental TensorFold/OMP.
🚀 1,081 tok/s peak aggregate
⚡ 791 tok/s average across the generation batch
🧠 Thinking ON, LOW effort
💾 64K-context
Local AI is getting wild @ashxhart and #TensorFold 🔥
What's new: compact GPU state, scheduling calibrated on measured multi-branch costs and faster short prompt-tail processing. Complete proposals return as files, with per-task token budgets, so the coordinator can integrate all twelve.
1
1
499




