Microsoft launches MAI-Code-1.1-Flash as a local model for Copilot
Understand this story
Microsoft compressed the 137.9B-parameter MoE coding model MAI-Code-1.1-Flash via mixed-precision quantization from about 276 GB at 16-bit to 53.0 GiB (3.31 bpw), and runs it as a local model for GitHub Copilot on RTX Spark PCs. The 53.0 GiB checkpoint has been released through Microsoft Foundry and is also offered as a local model for GitHub Copilot. Microsoft says RTX Spark PCs feature 128 GB of unified memory, allowing developers to run coding models on their own machines. Its performance comparisons are based on RTX Spark (128 GB unified memory, 64 GB GPU allocation), comparing the same Q2-XS target running alone versus paired with a DFlash2 draft across six input lengths from 1K to 120K tokens; Microsoft states these figures describe only that specific Q2-XS and DFlash2 configuration and should not be generalized to other serving conditions.
AI overview based on source reports · Updated 1 days ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIAdvancing the Quality-Efficiency Frontier for Coding AI on Windows PCs
Microsoft describes MAI-Code-1.1-Flash, a 137.9-billion-parameter mixture-of-experts coding model quantized from roughly 276 GB at 16-bit precision to a 53.0 GiB mixed-precision checkpoint (with a 46.1 GiB variant), running locally on NVIDIA RTX Spark PCs via GitHub Copilot. Using a DFlash2 speculative-decoding draft with layer-wise sliding-window attention, the measured RTX Spark configuration reached a median 51.12 tokens per second at 120K-token input context, with median end-to-end latency falling 27% from 527.25 to 382.41 seconds.
Activity for this story
- Current comparable activity
- 4
- Peak comparable activity
- 10Nov 5 06:00
- Change in the past 24 hours
- -50%
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.