Sakhanda Wire
NVDA $153.64 -1.60% MSFT $506.98 -2.13% GOOGL $279.95 -1.33% META $620.09 -1.01% AMZN $256.57 -0.46%
← Back to the news

Microsoft’s MAI Code 1.1 Flash trails DeepSeek in cost and benchmarks

Microsoft has released MAI Code 1.1 Flash, a coding model intended for GitHub Copilot. The company says it produces better code than its predecessor, uses 25 percent fewer tokens, and costs one quarter as much as the model released in June. Training used hundreds of thousands of reinforcement-learning environments in GitHub Copilot. Microsoft also reported that developers accepted 4 percent more of the model’s output.

In the published comparisons, MAI Code 1.1 Flash scored 72.6% on SWE-bench Verified and 62.9% on Terminal Bench 2.1. Those results were ahead of MAI Code 1 Flash, Claude Haiku 4.5 and GPT-5.4 mini, but below DeepSeek-V4-Flash-0731 on Terminal Bench, where DeepSeek scored 82.7%. A result for DeepSeek on SWE-bench Verified was not published.

The listed prices are $0.20 per million input tokens, $0.02 for cached input and $1.20 for output. DeepSeek-V4-Flash is listed at $0.14, $0.0028 and $0.28 respectively. Microsoft has not indicated that MAI Code 1.1 Flash will be released with open weights. The article links the model’s development to Microsoft’s effort to use cheaper in-house systems across Copilot.

Written by Sakhanda Wire AI from reporting by The Decoder. Facts, quotes and full context are in the original article below.

Originally published by The Decoder on

Read the original on The Decoder ↗

The reporting and the image belong to The Decoder and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news