Sign in
MAI-Code-1-Flash logo

MAI-Code-1-Flash

Microsoft · Released Jun 2026

View pricingCompare models
Conditional

MAI-Code-1-Flash was Microsoft's first in-house coding model, a sparse MoE priced at $0.75/$4.50 per 1M tokens. GitHub retired it from Copilot on 10 September 2026 and names MAI-Code-1.1-Flash, at $0.20/$1.20, as the replacement; it no longer appears in Copilot's supported-models or pricing tables. Copilot users should switch to 1.1-Flash, which is cheaper and scores higher on Terminal-Bench 2.1.

Is it right for you?

Good for

  • Historical comparison only: GitHub retired it from Copilot on 10 September 2026, so new Copilot work should use MAI-Code-1.1-Flash
  • Fast, iterative agentic coding workflows where latency and token cost matter more than frontier reasoning, on any host that still serves it

Not good for

  • New Copilot workflows: the model no longer appears in GitHub Copilot's supported-models or pricing tables after its 10 September 2026 retirement
  • Cost-sensitive teams: its successor MAI-Code-1.1-Flash lists at $0.20/$1.20 per 1M against this model's $0.75/$4.50
  • Complex multi-file architectural changes or novel algorithm design requiring frontier-level code reasoning
  • General-purpose non-coding tasks — this is a specialized coding model, not a general assistant

How it performs by task

Code completion and autocomplete

Excellent

Excellent — low latency, adaptive response length, purpose-built for Copilot's interactive loop

Code refactoring

Very Good

Strong — handles routine refactoring well; 60% fewer tokens than competitors on SWE-Bench Verified

Instruction following (coding)

Excellent

Best in class for lightweight tier — +28.9 points on IF Bench over Haiku 4.5; 85.8% on adversarial reasoning

Agentic tool use (Copilot harness)

Very Good

Strong — trained with Copilot's production harness; beats Haiku 4.5 on τ¹-Bench

Real-world bug fixes (SWE-Bench Pro)

Good

Solid for its size — 51.2% pass rate; 16-point lead over Haiku 4.5 but well below frontier models

Multi-file architecture / novel design

Fair

Weak — 5B active parameters cap reasoning depth; use a flagship model for complex work

Math and science reasoning

Very Good

Good — 92.5% AIME 2026, 84.6% GPQA Diamond; strong for a lightweight model

General-purpose chat / non-coding

Poor

Not designed for it — a specialized coding model; use GPT-5.5 or Claude for general tasks

Pricing

Input

$0.75 / 1M tokens

Output

$4.50 / 1M tokens

Context

Retired from Copilot Sep 10 2026; 256K ctx

View full pricing

Benchmarks

BenchmarkScoreSource
SWE-Bench Pro51.2% (vs Haiku 4.5: 35.2%) Source
SWE-Bench VerifiedBeats Haiku 4.5 with up to 60% fewer tokens Source
AIME 202692.5% (vs Haiku 4.5: 83.3%) Source
GPQA Diamond84.6% (vs Haiku 4.5: 73.2%) Source
IF Bench75.0 (vs Haiku 4.5: 46.1) Source
Adversarial reasoning85.8% adjusted accuracy Source
Frontier Math (Tier 1-3)6.3% (vs Haiku 4.5: 2.8%) Source
Frontier Science58.2% (vs Haiku 4.5: 42.3%) Source
HLE18.0% (vs Haiku 4.5: 9.5%) Source

Verdict history

Sep 26, 2026
conditional -> conditional (summary corrected: retired from Copilot): rating_before=conditional, rating_after=conditional. The summary still called it the model for high-volume Copilot workflows. GitHub's changelog set its deprecation for 2026-09-10, and on 2026-09-26 it is absent from Copilot's supported-models and pricing tables. The rating is held pending an operator ruling, since a retired model may merit caution or avoid.
Jul 15, 2026
copy edit to brevity standard: detail preserved in existing structured children

Sources

Verification log

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

  • Status· Updated

    Automated agent

    GitHub's changelog set MAI-Code-1-Flash's deprecation for 2026-09-10 and names MAI-Code-1.1-Flash as the replacement. On 2026-09-26 the model is absent from both the Copilot models-and-pricing table and the supported-models table, while MAI-Code-1.1-Flash is listed GA at $0.20/$1.20. Fixed via model_update: pricing_context is now 'Retired from Copilot Sep 10 2026; 256K ctx'; strengths and audiences were rewritten to drop the 'all Copilot tiers' and 'stretch Copilot credits' claims and to point to 1.1-Flash; the changelog and both GitHub docs pages were added as sources. The summary was corrected via verdict_set e9dd3180, rating_before=conditional and rating_after=conditional; the rating is held pending an operator ruling. Price scalars are unchanged. No draft post: the retirement was announced on 2026-08-11 and took effect 16 days ago.

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

    Pricing unchanged: $0.75/$4.50 per 1M tokens, 256K context. Confirmed live on GitHub Copilot pricing page.

  • Pricing· No changes

    Imported at launch

  • Profile· No changes

    Imported at launch

How we evaluate