Model A

Microsoft

52.27/100

Estimated · Public rank #123

90% interval 42.4–62.1

MAI-Thinking-1 vs Qwen3.7 Max

Updated September 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Model B

Qwen3.7 Max

Alibaba

68.56/100

Supported · Public rank #28

90% interval 62.5–74.6

Decision reading

Qwen3.7 Max has the higher public score, 68.56 versus 52.27, and the 90% score intervals do not overlap.

9 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Share or export

Share on X LinkedInSocial card CSV JSON

Which one for your work

Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.

Long documents
Prompts that approach the documented context limit
Qwen3.7 Max
Qwen3.7 Max has the larger documented context window.
Confidence: documented

Show secondary and unsupported calls

Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
MAI-Thinking-1 is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
MAI-Thinking-1 is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates

What is actually comparable

Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.

5 MAI-Thinking-1 only

9 Shared

32 Qwen3.7 Max only

Shared results: 9
MAI-Thinking-1 only: 5
Qwen3.7 Max only: 32
Like-for-like categories: 1 / 8

3 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.

Instruction following

Like-for-like

MAI-Thinking-1: 94.7; #1/120
Qwen3.7 Max: 91.1; #16/120
Basis: Provisional lane · 1 vs 1 weighted rows
Reading: MAI-Thinking-1 leads

Agentic

Directional only

MAI-Thinking-1: 51.7; Estimated · #53/151
Qwen3.7 Max: 42.8; Supported · #110/151
Basis: BenchAlign lane · 1 vs 10 public rows
Reading: Directional only

Coding

Directional only

MAI-Thinking-1: 51.8; Estimated · #60/183
Qwen3.7 Max: 49.5; Supported · #77/183
Basis: BenchAlign lane · 4 vs 10 public rows
Reading: Directional only

Knowledge

Directional only

MAI-Thinking-1: 53.3; Estimated · #68/181
Qwen3.7 Max: 62.4; Supported · #28/181
Basis: BenchAlign lane · 4 vs 9 public rows
Reading: Directional only

Reasoning

Not comparable

MAI-Thinking-1: Not ranked
Qwen3.7 Max: 74.8; Unranked · 3 rankable rows
Basis: Provisional lane · 0 vs 1 weighted rows
Reading: Not comparable

Math

Not comparable

MAI-Thinking-1: 73.4; Unranked · 3 rankable rows
Qwen3.7 Max: 82.1; Unranked · 3 rankable rows
Basis: Provisional lane · 2 vs 1 weighted rows
Reading: Not comparable

Multilingual

Not comparable

MAI-Thinking-1: Not ranked
Qwen3.7 Max: 100.0; #1/12
Basis: Provisional lane · 0 vs 1 weighted rows
Reading: Not comparable

Multimodal

Not comparable

MAI-Thinking-1: Not ranked
Qwen3.7 Max: Not ranked
Basis: Provisional lane · 0 vs 0 weighted rows
Reading: Not comparable

Category scores with the server-provided evidence basis for MAI-Thinking-1 and Qwen3.7 Max
Category	MAI-Thinking-1	Qwen3.7 Max	Basis	Reading
Instruction following	94.7#1/120	91.1#16/120	Like-for-likeProvisional lane · 1 vs 1 weighted rows	MAI-Thinking-1 leads
Agentic	51.7Estimated · #53/151	42.8Supported · #110/151	Directional onlyBenchAlign lane · 1 vs 10 public rows	Directional only
Coding	51.8Estimated · #60/183	49.5Supported · #77/183	Directional onlyBenchAlign lane · 4 vs 10 public rows	Directional only
Knowledge	53.3Estimated · #68/181	62.4Supported · #28/181	Directional onlyBenchAlign lane · 4 vs 9 public rows	Directional only
Reasoning	Not ranked	74.8Unranked · 3 rankable rows	Not comparableProvisional lane · 0 vs 1 weighted rows	Not comparable
Math	73.4Unranked · 3 rankable rows	82.1Unranked · 3 rankable rows	Not comparableProvisional lane · 2 vs 1 weighted rows	Not comparable
Multilingual	Not ranked	100.0#1/12	Not comparableProvisional lane · 0 vs 1 weighted rows	Not comparable
Multimodal	Not ranked	Not ranked	Not comparableProvisional lane · 0 vs 0 weighted rows	Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.

Shape of the matched evidence

Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.

Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.

Terminal-Bench 2.0
Agentic
MAI-Thinking-1: 46%Qwen3.7 Max: 69.7%Normalized gap 23.7MAI-Thinking-1 source Qwen3.7 Max source
HMMT Feb 2026
Math
MAI-Thinking-1: 84.9%Qwen3.7 Max: 97.1%Normalized gap 12.2MAI-Thinking-1 source Qwen3.7 Max source
GPQA
Knowledge
MAI-Thinking-1: 84.2%Qwen3.7 Max: 92.4%Normalized gap 8.2MAI-Thinking-1 source Qwen3.7 Max source
SWE-bench Pro
Coding
MAI-Thinking-1: 52.8%Qwen3.7 Max: 60.6%Normalized gap 7.8MAI-Thinking-1 source Qwen3.7 Max source
SWE-bench Verified
Coding
MAI-Thinking-1: 73.5%Qwen3.7 Max: 80.4%Normalized gap 6.9MAI-Thinking-1 source Qwen3.7 Max source

What each workload costs

Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.

Chat turn

1K fresh input + 500 output tokens

MAI-Thinking-1: API rate not published; Fits in one request
Qwen3.7 Max: API rate not published; Fits in one request

MAI-Thinking-1 has no comparable published API token rate. Qwen3.7 Max has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

MAI-Thinking-1: API rate not published; Fits in one request
Qwen3.7 Max: API rate not published; Fits in one request

MAI-Thinking-1 has no comparable published API token rate. Qwen3.7 Max has no comparable published API token rate.

Cache-heavy agent loop

200K cached + 20K fresh input + 10K output tokens

MAI-Thinking-1: API rate not published; Fits in one request; Cached-input rate unavailable
Qwen3.7 Max: API rate not published; Fits in one request; Cached-input rate unavailable

MAI-Thinking-1 has no comparable published API token rate. Qwen3.7 Max has no comparable published API token rate.

Specification differences

Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.

SpecificationMAI-Thinking-1Qwen3.7 Max

Context window

Maximum documented context; output-token limits may be lower.

MAI-Thinking-1

256K

Qwen3.7 Max

API model ID

MAI-Thinking-1

Not sourced

Qwen3.7 Max

Not sourced

Cached-input rate

A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.

MAI-Thinking-1

No comparable hosted API rate

Qwen3.7 Max

No comparable hosted API rate

Documented inputs

MAI-Thinking-1

Not sourced

Qwen3.7 Max

Not sourced

Documented outputs

MAI-Thinking-1

Not sourced

Qwen3.7 Max

Not sourced

Provider availability

MAI-Thinking-1

Not sourced

Qwen3.7 Max

Not sourced

Reasoning profile

MAI-Thinking-1

Reasoning

Qwen3.7 Max

Reasoning

Weight access

MAI-Thinking-1

Proprietary

Qwen3.7 Max

Proprietary

License

MAI-Thinking-1

Proprietary

Qwen3.7 Max

Proprietary

Release date

MAI-Thinking-1

2026-06-02

Qwen3.7 Max

2026-05-16

If you already use one of these models

Deployment change: The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
Quality signal: Qwen3.7 Max has the higher public score, 68.56 versus 52.27, and the 90% score intervals do not overlap.
Workload cost: A complete comparable API-rate estimate is not available for both models.
Context tradeoff: Qwen3.7 Max has the larger documented window (1M).

Run the same representative tasks against both endpoints before changing production traffic.

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence46 rows

Agentic

Terminal-Bench 2.0
MAI-Thinking-146%
Source
Qwen3.7 Max69.7%
Source
Qwen3.7 Max leads this result
QwenClawBench
MAI-Thinking-1—
Qwen3.7 Max64.3%
Source
Not directly comparable
Claw-Eval
MAI-Thinking-1—
Qwen3.7 Max65.2%
Source
Not directly comparable
BFCL v4
MAI-Thinking-1—
Qwen3.7 Max75.0%
Source
Not directly comparable
MCP Atlas
MAI-Thinking-1—
Qwen3.7 Max76.4%
Source
Not directly comparable
VITA-Bench
MAI-Thinking-1—
Qwen3.7 Max47.9%
Source
Not directly comparable
HLE w/ tools
MAI-Thinking-1—
Qwen3.7 Max53.5%
Source
Not directly comparable
Gert Labs
MAI-Thinking-1—
Qwen3.7 Max64.27%
Source
Not directly comparable
ResearchClawBench
MAI-Thinking-1—
Qwen3.7 Max18.7%
Source
Not directly comparable
Terminal-Bench 2.1 (Vals)
MAI-Thinking-1—
Qwen3.7 Max61.0%
Source
Not directly comparable

Coding

LiveCodeBench v6
MAI-Thinking-187.7%
Source
Qwen3.7 Max—
Not directly comparable
SWE-bench Verified
MAI-Thinking-173.5%
Source
Qwen3.7 Max80.4%
Source
Qwen3.7 Max leads this result
SWE-bench Pro
MAI-Thinking-152.8%
Source
Qwen3.7 Max60.6%
Source
Qwen3.7 Max leads this result
Terminal-Bench 2.0
MAI-Thinking-146.0%
Source
Qwen3.7 Max69.7%
Source
Qwen3.7 Max leads this result
SWE Multilingual
MAI-Thinking-1—
Qwen3.7 Max78.3%
Source
Not directly comparable
NL2Repo
MAI-Thinking-1—
Qwen3.7 Max47.2%
Source
Not directly comparable
SciCode
MAI-Thinking-1—
Qwen3.7 Max53.5%
Source
Not directly comparable
LiveCodeBench
MAI-Thinking-1—
Qwen3.7 Max91.6%
Source
Not directly comparable
OpenHarmony Bench
MAI-Thinking-1—
Qwen3.7 Max53.4%
Source
Not directly comparable
LiveCodeBench (Vals)
MAI-Thinking-1—
Qwen3.7 Max87.1%
Source
Not directly comparable
SWE-bench (Vals)
MAI-Thinking-1—
Qwen3.7 Max68.8%
Source
Not directly comparable

Reasoning

Graphwalks BFS 128K
MAI-Thinking-190%
Source
Qwen3.7 Max—
Not directly comparable
MRCRv2
MAI-Thinking-1—
Qwen3.7 Max90.4%
Source
Not directly comparable
CritPt
MAI-Thinking-1—
Qwen3.7 Max13.4%
Source
Not directly comparable

Knowledge

GPQA
MAI-Thinking-184.2%
Source
Qwen3.7 Max92.4%
Source
Qwen3.7 Max leads this result
GPQA-D
MAI-Thinking-184.2%
Source
Qwen3.7 Max92.4%
Source
Qwen3.7 Max leads this result
MMLU-Pro
MAI-Thinking-185%
Source
Qwen3.7 Max89.6%
Source
Qwen3.7 Max leads this result
SimpleQA
MAI-Thinking-131%
Source
Qwen3.7 Max—
Not directly comparable
HLE
MAI-Thinking-1—
Qwen3.7 Max41.4%
Source
Not directly comparable
MMLU-Redux
MAI-Thinking-1—
Qwen3.7 Max95%
Source
Not directly comparable
SuperGPQA
MAI-Thinking-1—
Qwen3.7 Max73.6%
Source
Not directly comparable
MMMLU
MAI-Thinking-1—
Qwen3.7 Max90.3%
Source
Not directly comparable
GPQA Diamond (Vals)
MAI-Thinking-1—
Qwen3.7 Max90.2%
Source
Not directly comparable
MMLU-Pro (Vals)
MAI-Thinking-1—
Qwen3.7 Max89.3%
Source
Not directly comparable

Math

AIME 2025
MAI-Thinking-197%
Source
Qwen3.7 Max—
Not directly comparable
AIME26
MAI-Thinking-194.5%
Source
Qwen3.7 Max—
Not directly comparable
HMMT Feb 2026
MAI-Thinking-184.9%
Source
Qwen3.7 Max97.1%
Source
Qwen3.7 Max leads this result
IMOAnswerBench
MAI-Thinking-1—
Qwen3.7 Max90.0%
Source
Not directly comparable
Apex
MAI-Thinking-1—
Qwen3.7 Max44.5%
Source
Not directly comparable

Multilingual

MMLU-ProX
MAI-Thinking-1—
Qwen3.7 Max87%
Source
Not directly comparable
NOVA-63
MAI-Thinking-1—
Qwen3.7 Max59.0%
Source
Not directly comparable
INCLUDE
MAI-Thinking-1—
Qwen3.7 Max86.2%
Source
Not directly comparable
MAXIFE
MAI-Thinking-1—
Qwen3.7 Max89.2%
Source
Not directly comparable
PolyMath
MAI-Thinking-1—
Qwen3.7 Max86.5%
Source
Not directly comparable

Instruction following

IFBench
MAI-Thinking-185%
Source
Qwen3.7 Max79.1%
Source
MAI-Thinking-1 leads this result
IFEval
MAI-Thinking-1—
Qwen3.7 Max94.3%
Source
Not directly comparable

Frequently asked questions

Which is better, MAI-Thinking-1 or Qwen3.7 Max?

Qwen3.7 Max has the higher public score, 68.56 versus 52.27, and the 90% score intervals do not overlap. That is the clearest overall quality signal in the current public data.

Which is better for coding, MAI-Thinking-1 or Qwen3.7 Max?

MAI-Thinking-1 scores higher for coding on the public lane, 51.8 to 49.5. MAI-Thinking-1 is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which is better for agentic tasks, MAI-Thinking-1 or Qwen3.7 Max?

MAI-Thinking-1 scores higher for agentic tasks on the public lane, 51.7 to 42.8. MAI-Thinking-1 is scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which costs less, MAI-Thinking-1 or Qwen3.7 Max?

Both models do not have comparable published API token rates, so this page does not name a universal price winner.

Which has the larger context window, MAI-Thinking-1 or Qwen3.7 Max?

Qwen3.7 Max has the larger documented context window: 1M, compared with 256K.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated September 4, 2026

Watch MAI-Thinking-1 vs Qwen3.7 Max

One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.

Read a sample issue

Join 2,000+ readers.

MAI-Thinking-1 vs Qwen3.7 Max

Which one for your work

Long documents

Coding work

Agentic work

Chat turn cost

Cache-heavy agent loop cost

Repository review cost

What is actually comparable

Category results, on a stated basis

Instruction following

Agentic

Coding

Knowledge

Reasoning

Math

Multilingual

Multimodal

Shape of the matched evidence

What each workload costs

Chat turn

Repository review

Cache-heavy agent loop

Specification differences

Context window

API model ID

Cached-input rate

Documented inputs

Documented outputs

Provider availability

Reasoning profile

Weight access

License

Release date

Benchmark evidence

Agentic

Coding

Reasoning

Knowledge

Math

Multilingual

Instruction following

Frequently asked questions

Which is better, MAI-Thinking-1 or Qwen3.7 Max?

Which is better for coding, MAI-Thinking-1 or Qwen3.7 Max?

Which is better for agentic tasks, MAI-Thinking-1 or Qwen3.7 Max?

Which costs less, MAI-Thinking-1 or Qwen3.7 Max?

Which has the larger context window, MAI-Thinking-1 or Qwen3.7 Max?

Related comparisons

Watch MAI-Thinking-1 vs Qwen3.7 Max