MAI Code 2 and MAI Code 2 Flash by Microsoft added to BenchGecko
Microsoft's new coding models · 1,050,000 token context · MAI Code 2 at $1 input and $5 output per million tokens, Flash at $0.10 and $0.50, both 50% off until May 17, 2027.
What was added
Microsoft released MAI Code 2 and MAI Code 2 Flash on November 17, 2026, as part of seven MAI models announced at Ignite. Both are now listed on BenchGecko under Microsoft. They are coding models trained from scratch by Microsoft AI. MAI Code 2 targets complex work in real repositories, and MAI Code 2 Flash targets high volume tasks such as code explanations, targeted edits and routine fixes.
Context and access
Microsoft's model card lists a 1,050,000 token context length for both models, a maximum input of 922,000 tokens and a maximum output of 128,000 tokens. Both accept text and image input and return text, with selectable reasoning effort. Both are sparse Mixture of Experts models with an integrated vision encoder, and weights are not distributed.
Both models are available in GitHub Copilot. MAI Code 2 Flash is also in a limited access preview in Microsoft Foundry through an OpenAI compatible Responses API. Microsoft says the models will come to OpenRouter, Vercel, Hermes, OpenCode and Kilo, without a date.
Price on BenchGecko
BenchGecko lists the standard per token price from GitHub's Copilot pricing table, per million tokens:
- MAI Code 2 · $1 input · $5 output · $0.05 cached input
- MAI Code 2 Flash · $0.10 input · $0.50 output · $0.01 cached input
Until May 17, 2027 both run at a promotional 50% off: $0.50 input, $2.50 output and $0.025 cached input for MAI Code 2, and $0.05 input, $0.25 output and $0.005 cached input for Flash.
How it compares on price
At the standard price, MAI Code 2 costs half of GPT-6.1 Sol ($2 input, $10 output on BenchGecko) and 25% of Claude Opus 5.5 ($4 input, $20 output). MAI Code 2 Flash matches GPT-6 Luna and Claude Haiku 5.5 at $0.10 input and $0.50 output. During the promotion each price is half of those figures.
Vendor benchmarks
Microsoft reports that, averaged over DeepSWE 1.1, Terminal Bench 2.1, SWE-Atlas and FrontierCode v1.1, MAI Code 2 reaches 98% of GPT-6.1 Sol's score at 40% of the cost and 96% of Claude Opus 5.5's score at 5% of the cost. In a blind frontend evaluation with 258 annotators, reviewers preferred MAI Code 2 over GPT-6 Sol 64% of the time. Microsoft says MAI Code 2 Flash performs comparably to GPT-6 Luna and Claude Haiku 5.5 at 65% and 20% of their cost. These cost figures use the promotional price. They are self reported, and BenchGecko has no independent scores yet.
The other Ignite models cover cyber defense, images, speech, transcription and routing and are not added as language model rows for now.
Sources
Source: BenchGecko
Written automatically from BenchGecko's tracked data for this period; every number comes from the changelog, model, mindshare and pricing datasets. Checked against the source data before publishing.