Skip to main content
AI Choice EngineAI Choice Engine

Model comparison

Microsoft MAI-Thinking-1 vs Qwen3-Max

Microsoft against Qwen, compared on context, price, and verified benchmark results.

Catalog record checked September 5, 2026Individual provider fields may changeEvidence confidence: High — see receipts below

Microsoft

Microsoft MAI-Thinking-1

Frontier

vs

Qwen

Qwen3-Max

Frontier

AI model capability comparison
SpecificationMicrosoft MAI-Thinking-1Qwen3-Max
ProviderMicrosoftQwen
TierFrontierFrontier
Context window256KWinner: 262K
Max output64KWinner: 66K
Input / 1M tokens$2Winner: $1.20
Output / 1M tokens$8Winner: $6
WeightsClosedClosed
Parameters~1T total / 35B active (sparse MoE)1T (proprietary)
Reasoning levelslow, medium, highlow, high, max
Modalitiestexttext, image, video
API model idmai-thinking-1qwen3-max
ReleasedJune 2, 2026September 23, 2025
Artificial Analysis Intelligence Index (2026-08-14)Winner: 5552
AIME 2025 (2026-08-12)97Not verifiedUnverified
SWE-bench Verified (2026-06)73.5Not verifiedUnverified
GPQA Diamond (2026-06)84.2Not verifiedUnverified
Terminal-Bench 2.0 (2026-06)46Not verifiedUnverified

Prices are USD per million tokens at standard rates, excluding batch and caching discounts. Bold indicates the better figure where one is objectively better. Values we could not confirm from the provider are shown as “Not verified” rather than estimated.

Benchmark receipts

  • 2026-08-14: Artificial AnalysisComposite index blending reasoning, knowledge, and coding evals into one 0–100 score. Comparability: directly comparable — AA occasionally rebaselines the index scale between snapshots — a score captured on one date is only comparable to same-snapshot scores (check measuredAt).
  • 2026-08-12: Microsoft AI — Introducing MAI-Thinking-1 (vendor, 256k output)Competition mathematics: AIME 2025 exam problems, typically pass@1 with tools disallowed or single-attempt code execution per the harness. Comparability: comparable with caveat — 30-question total denominator: one problem = ~3.3 points. Tool-use policy (calculator/code interpreter) must match between compared models.
  • 2026-06: Microsoft AI model page (vendor, as labeled: Terminal-Bench 2.0 — NOT comparable to 2.1 observations)Real GitHub issue resolution: does the model's patch pass the hidden tests. Comparability: comparable with caveat — Post-audit vendor claims and pre-audit scores sit on different task trust levels; scaffolding (agent harness, compute budget) also dominates results. Never aggregate across scaffolds.

Pricing tiers: Microsoft MAI-Thinking-1: Global Standard $2.00/$8.00 per MTok, cached input $0.20 (Azure Retail Prices API, effective 2026-08-01, uniform across ~22 public regions; US Gov regions carry premiums). Still preview; Global Standard only, no PTU yet. 256K context with a 64K output cap that includes reasoning tokens. ~1T/35B sparse MoE; reasoning is always on (no effort levels). Microsoft's first true reasoning model, trained without OpenAI distillation. · Qwen3-Max: Qwen API $1.20/$6.00 per MTok (≤32K); higher tiers for 32K–256K. OpenRouter lists ~$0.78/$3.90. Proprietary, API-only.

FrontierRecord checked September 5, 2026

Microsoft MAI-Thinking-1

MAI-Thinking-1 is Microsoft's first in-house reasoning model, reducing dependence on OpenAI.

Best for

  • Enterprise (Azure)
  • Reasoning
  • Microsoft ecosystem

Watch out

Foundry-gated preview; no MAI-2 exists yet. Of the Build 2026 MAI wave, only MAI-Thinking-1 and MAI-Cyber-1-Flash ($0.60/$3.50, released 2026-07-27) have published token meters.

FrontierRecord checked September 3, 2026

Qwen3-Max

Qwen3-Max is Alibaba's proprietary 1-trillion-parameter API model with multimodal input.

Best for

  • Agentic coding
  • Multimodal knowledge work
  • Long-context via API

Watch out

Max line is proprietary/API-only (no published weights). Per-provider prices differ.

When the cheaper one wins

Qwen3-Max is cheaper on output at $6 per million tokens against $8 for Microsoft MAI-Thinking-1 — about 1.3×. Use the cheaper tier for classification, extraction, summarisation, and any task where the expensive model’s extra score does not change the accepted output. The expensive one only pays if your hardest task actually fails on the cheap tier. These are standard-tier API rates, excluding batch and cache discounts.

Run the model picker

Evidence confidence: High

How strong and complete the evidence behind this comparison is — not a prediction of which model is better.

  • Pricing verified on both sidesInput and output rates are verified for both models.
  • 5/5 core specs verified on both sidesAll core specifications verified for both models.
  • 1 shared named benchmark with differing scoresMeasured on: Artificial Analysis Intelligence Index.
  • Verified within the last 90 daysNewest catalog check was 6 days ago.
  • Both models carry source citationsEach side has at least two catalog sources on record.

Source receipts

Each catalog figure was checked against the provider or an independent second source on the date shown.

Common questions

Microsoft MAI-Thinking-1 vs Qwen3-Max

Answered from the verified figures on this page rather than general guidance.

Is Microsoft MAI-Thinking-1 or Qwen3-Max cheaper for input?

Qwen3-Max is cheaper at $1.20 per million input tokens, against $2 for Microsoft MAI-Thinking-1 — roughly 1.7× the price. Output tokens usually dominate a real bill, so weigh the output rate more heavily than the input rate. These are base rates; Microsoft MAI-Thinking-1 has tiered pricing: Global Standard $2.00/$8.00 per MTok, cached input $0.20 (Azure Retail Prices API, effective 2026-08-01, uniform across ~22 public regions; US Gov regions carry premiums). Still preview; Global Standard only, no PTU yet. 256K context with a 64K output cap that includes reasoning tokens. ~1T/35B sparse MoE; reasoning is always on (no effort levels). Microsoft's first true reasoning model, trained without OpenAI distillation. Qwen3-Max has tiered pricing: Qwen API $1.20/$6.00 per MTok (≤32K); higher tiers for 32K–256K. OpenRouter lists ~$0.78/$3.90. Proprietary, API-only.

Is Microsoft MAI-Thinking-1 or Qwen3-Max cheaper for output?

Qwen3-Max is cheaper at $6 per million output tokens, against $8 for Microsoft MAI-Thinking-1 — roughly 1.3× the price. Output tokens usually dominate a real bill, so weigh the output rate more heavily than the input rate. These are base rates; Microsoft MAI-Thinking-1 has tiered pricing: Global Standard $2.00/$8.00 per MTok, cached input $0.20 (Azure Retail Prices API, effective 2026-08-01, uniform across ~22 public regions; US Gov regions carry premiums). Still preview; Global Standard only, no PTU yet. 256K context with a 64K output cap that includes reasoning tokens. ~1T/35B sparse MoE; reasoning is always on (no effort levels). Microsoft's first true reasoning model, trained without OpenAI distillation. Qwen3-Max has tiered pricing: Qwen API $1.20/$6.00 per MTok (≤32K); higher tiers for 32K–256K. OpenRouter lists ~$0.78/$3.90. Proprietary, API-only.

Which has the larger context window, Microsoft MAI-Thinking-1 or Qwen3-Max?

Qwen3-Max accepts 262K tokens against 256K for Microsoft MAI-Thinking-1. This only matters if you routinely send very long documents or large codebases.

Do Microsoft MAI-Thinking-1 and Qwen3-Max support the same reasoning levels?

Microsoft MAI-Thinking-1 exposes low, medium, high, while Qwen3-Max exposes low, high, max.

Should I use Microsoft MAI-Thinking-1 or Qwen3-Max?

Both sit in the frontier tier, so the choice usually comes down to price and context rather than capability. Microsoft MAI-Thinking-1 suits enterprise (azure); Qwen3-Max suits agentic coding.

Next step

Choosing between them

The verified figures that separate this pair, computed from the catalog rather than restated boilerplate.

  • Input price: Qwen3-Max costs $1.20 per 1M tokens versus $2 for Microsoft MAI-Thinking-1 — a 1.7x difference at the headline tier.
  • Context: Qwen3-Max takes 262K against 256K for Microsoft MAI-Thinking-1 — only decisive if your prompts approach the smaller window.
  • Measured capability: Microsoft MAI-Thinking-1 leads Artificial Analysis Intelligence Index 55 to 52 (measured 2026-08-14).

Choosing a harness rather than a model — Cursor, Copilot, Claude Code, Windsurf, or Lovable? Compare agentic harnesses · Latest releases.