Stealth Local AI Startup published PR 954 for DeepSeek V4 Flash with a 4% prefill and 6.5% decode.
Public source
Publisher name
Public post
My contribution to DwarfStar is continuing! I use it in my "Production" AI Lab so I keep pushing it. - PR 953: GLM 5.3 Flash +3.5% prefill and improved mathematical accu…
Company
Stealth Local AI Startup
Building something for the local AI community.
- Industry
- Technology, Information and Internet
- Company size
- 1 employee
About Stealth Local AI Startup
Building something for the local AI community.
See moreLatest activity
Latest activity from Stealth Local AI Startup
4 signals
Products & Services
Stealth Local AI Startup announced that Fable 5.1 will be the entry-level of Local AI models.
Products & Services
Stealth Local AI Startup published PR 953 for GLM 5.3 Flash with a 3.5% prefill and improved mathematical accuracy on the M5 (TensorOps) kernel.
Products & Services
Stealth Local AI Startup is preparing for upcoming support for Qwen 3.8 Flash Next model following Antirez suggestions on the kernel architecture.
Discover more
Similar signals
Similar public activity from other companies.
Products & Services
Stealth AI Startup
Stealth AI Startup announced that Astra is an extremely smart model with a cost that is way less than others.
Products & Services
Unsloth AI
Unsloth AI made GLM-5.3-Flash run 3.3x faster locally with optimized decoding and bonus multi-token prediction support.
Products & Services
Yantrion Inc
Yantrion Inc is starting to turn GPU memory savings into more tokens per second on the same hardware.
Products & Services
turbopuffer
turbopuffer achieves 17ms p50 hybrid latency
Products & Services
Stealth Startup