MAI-Code-1.1-Flash: Microsoft's Model in Copilot
GitHub Copilot added Microsoft's MAI-Code-1.1-Flash on 11 August 2026, priced 73% below its predecessor, as Microsoft reduces reliance on OpenAI models.
GitHub made Microsoft's MAI-Code-1.1-Flash model available inside GitHub Copilot on 11 August 2026, according to GitHub's own changelog [1]. The model is billed as "Microsoft's latest small-tier coding model," adds native vision support for image understanding, and carries a list price 73% lower than its predecessor, MAI-Code-1-Flash [1]. It is the second in-house Microsoft coding model to reach Copilot inside three months: MAI-Code-1-Flash itself first became available in Copilot on 2 June 2026, following its unveiling at Microsoft's Build 2026 developer conference [2].
Copilot has offered models from multiple vendors — OpenAI, Anthropic and, as of 12 August 2026, xAI's Grok 4.6 — for some time. MAI-Code-1.1-Flash is different in one respect: it's Microsoft's own model, built in-house rather than licensed from a partner.
What GitHub actually announced
According to GitHub's changelog, MAI-Code-1.1-Flash brings "improvements in coding quality, instruction following, tool use, and performance" over its predecessor, alongside the new native vision capability [1]. On pricing, GitHub states the model runs at a 0.25× premium request multiplier for annual Copilot subscribers, and at standard usage-based billing under provider list pricing otherwise [1] — mechanically similar to how Copilot already prices other included models, but at a lower multiplier than MAI-Code-1-Flash carried.
Where it's available, and where it isn't yet
GitHub lists availability for Free and Student users through automatic model selection only, while Pro, Pro+, Max, Business and Enterprise subscribers can select it manually as well [1]. The model reaches Copilot CLI, the cloud agent, the GitHub Copilot app, Copilot Chat on GitHub, VS Code, Visual Studio, GitHub Mobile, JetBrains IDEs, Eclipse and Xcode [1]. For Business and Enterprise customers specifically, it is off by default: administrators must manually enable the MAI-Code-1.1-Flash policy in Copilot settings before their users can access it [1].
Why this matters
Every model Copilot has offered until this year has come from an external partner — primarily OpenAI, with Anthropic's Claude and now xAI's Grok 4.6 added over 2026. MAI-Code-1.1-Flash is Microsoft using its own "MAI" (Microsoft AI) model family inside its own developer product, which matters regardless of how the model performs: it reduces Microsoft's dependence on any single external model provider for a product it doesn't fully control the roadmap or pricing of. A 73% price cut from the first MAI-Code-1-Flash version to this one also suggests Microsoft is iterating on cost efficiency quickly, which is a meaningful signal for a company that pays those partner-model costs at enormous scale across Copilot's user base.
The timing also sits inside a broader pattern across the industry in mid-2026: OpenAI split GPT-5.6 into three cost-differentiated tiers on 9 July 2026, and xAI priced Grok 4.6 aggressively below both OpenAI's and Anthropic's standard rates when it launched on 12 August 2026. Every major AI lab is now visibly competing on cost-efficient, smaller-tier models as much as on flagship capability, and Microsoft building its own small-tier model in-house — rather than simply reselling a partner's cheapest tier — fits that same direction.
Who should care
Individual developers on Copilot's Free or Student tiers will encounter MAI-Code-1.1-Flash automatically through model auto-selection without necessarily choosing it, so it's worth knowing what's actually running your completions and chat. Engineering leaders managing Copilot Business or Enterprise seats need to actively decide whether to enable it, since it's opt-in for those tiers — a decision worth making deliberately rather than leaving unaddressed. Anyone comparing coding-assistant costs at scale should note the pricing multiplier difference between MAI-Code-1.1-Flash and Copilot's OpenAI- or Anthropic-backed options, since a smaller, cheaper model may be entirely adequate for routine completions even if it isn't the strongest available option for complex reasoning.
Practical implications for buyers and users
Because MAI-Code-1.1-Flash is explicitly positioned as a "small-tier" model, the realistic comparison isn't against GPT-5.6 Sol or Claude Opus 5 for hard problems, but against other fast, low-cost coding models for routine, high-volume completion work. The new vision support is worth testing directly if your workflow involves screenshots, diagrams or UI mockups as coding context, since that's a genuinely new capability rather than a performance tweak. Enterprise administrators should treat the default-off policy setting as a deliberate gate to evaluate the model against their existing choices before enabling it fleet-wide, rather than something to switch on reflexively.
Limitations, availability and unresolved questions
GitHub's changelog does not publish independent benchmark comparisons between MAI-Code-1.1-Flash and Copilot's other available models, so how it actually performs against OpenAI's or Anthropic's small-tier offerings on real coding tasks is not established in the material we reviewed. Nor does GitHub disclose the model's underlying architecture, parameter count or training data — consistent with how model cards for Copilot's other options are typically handled, but a limitation for anyone trying to reason about the model's likely strengths and weaknesses. It's also not yet clear whether Microsoft plans further MAI-Code releases on a similar roughly two-month cadence to the one seen between the original MAI-Code-1-Flash (2 June 2026) and this update (11 August 2026), or whether Microsoft intends to eventually offer a larger, higher-capability MAI-Code tier alongside this small-tier line rather than continuing to rely on OpenAI and Anthropic models for Copilot's hardest coding tasks.
Verdict
MAI-Code-1.1-Flash is a modest but strategically telling update: a cheaper, vision-capable refresh of Microsoft's own in-house coding model, arriving roughly ten weeks after the first version. The pricing improvement is concrete and confirmed, and the multi-IDE availability means most Copilot users can try it without changing their existing workflow. Its real significance isn't raw capability — GitHub itself frames it as a small-tier model — but what it signals about Microsoft's direction: building and iterating on its own coding models inside Copilot, rather than remaining solely dependent on OpenAI's.
The current AI Writing & Research shortlist
Where this sits in the wider market: our current shortlist for AI Writing & Research, what each tool is best at and the main caution to check before committing.
| Tool | Best for | Current position | Important caution |
|---|---|---|---|
| ChatGPT Best all-rounder | General writing, analysis and multimodal work | GPT-5.6 combines strong reasoning with files, images, tools and broad workflow support. It is the safest starting point when one assistant must cover many jobs. | Teams should define data-handling rules and verify important claims. |
| Claude Long-form pick | Editorial work, complex documents and careful reasoning | Claude’s current Opus and Sonnet 5 family is built for sustained professional and agentic work, with a strong reputation for readable long-form output. | The highest-capability tiers can be unnecessary for routine copy. |
| Gemini Google ecosystem | Workspace users and multimodal source material | Gemini 3.7 Flash, documented in August 2026, is the current Flash release, connecting reasoning, multimodal inputs and Google’s productivity ecosystem. | Feature availability varies by Workspace plan and region. |
| Perplexity Research pick | Fast web research and cited discovery | Perplexity is useful when the first requirement is finding and comparing live web sources rather than drafting from memory. | A citation does not guarantee that the source supports every sentence; open the evidence. |
| Jasper Brand governance | Marketing teams with repeatable brand workflows | Jasper focuses on governed marketing content, brand context and campaign production rather than being a general-purpose chatbot. | Plans, limits and model availability change frequently; confirm the current vendor page before purchasing. |
| Copy.ai GTM workflows | Sales and marketing process automation | Copy.ai has evolved from a copy generator into a go-to-market workflow platform for repeatable content and sales operations. | Plans, limits and model availability change frequently; confirm the current vendor page before purchasing. |
| Writesonic AI visibility | SEO content and answer-engine monitoring | Writesonic combines assisted content production with tooling aimed at search and AI-answer visibility. | Plans, limits and model availability change frequently; confirm the current vendor page before purchasing. |
| Grammarly Editing layer | Everyday rewriting, tone and quality control | Grammarly works best as an editing and communication layer across existing applications rather than as the only writing system. | Plans, limits and model availability change frequently; confirm the current vendor page before purchasing. |
| Notion AI Knowledge workspace | Teams whose documents and projects already live in Notion | Notion AI is strongest when it can work inside an existing team knowledge base instead of requiring constant copying between tools. | Plans, limits and model availability change frequently; confirm the current vendor page before purchasing. |
| KoalaWriter SEO drafts | Structured long-form drafts and niche publishing | KoalaWriter remains a focused option for producing structured, search-aware drafts quickly. | Human research, original experience and fact-checking are still required before publishing. |
Related reading
Sources and verification notes
Primary product documentation checked for this update: