Model Details
Qwen3.7 Max
Quick answer: Qwen3.7 Max is Alibaba's March 2026 API-only model scoring 89.6% MMLU-Pro, 80.4% SWE-Bench Verified, 41.4% HLE, and 29.6% LHTB. Proprietary API.
At a Glance
Where Qwen3.7 Max leads
- 89.6% MMLU-Pro — excellent academic breadth
- 80.4% SWE-Bench Verified — strong software engineering
- Strong coding + knowledge combined profile
- March 2026: latest Qwen generation
Where it lags
- 41.4% HLE — moderate frontier exam reasoning
- 31.1% Agents Last Exam — limited complex agentic tasks
- Proprietary API: no open weights
Best for: SWE-Bench Verified focused coding pipelines; high-MMLU-Pro knowledge tasks; Alibaba Cloud Qwen API deployments.
What Qwen3.7 Max Is
Qwen3.7 Max is Alibaba's March 2026 proprietary API model — the "Max" tier of the Qwen3.7 generation, representing Alibaba's frontier offering at that release. The 80.4% SWE-Bench Verified is particularly strong for a March 2026 model.
Specifications
| Field | Value |
|---|---|
| Organization | Alibaba |
| License | Proprietary (API only) |
| Release date | March 2026 |
| Modality | Text only |
Pricing
Available via Alibaba Cloud Qwen API. Refer to Alibaba Cloud pricing.
Public Benchmark Scores
| Benchmark | Score | Source | Date |
|---|---|---|---|
| MMLU-Pro | 89.6% | Benchgen evaluation | 2026-03 |
| SWE-Bench Verified | 80.4% | Benchgen evaluation | 2026-03 |
| Humanity's Last Exam | 41.4% | Benchgen evaluation | 2026-03 |
| Agents Last Exam | 31.1% | Benchgen evaluation | 2026-03 |
| LHTB | 29.6% | Benchgen evaluation | 2026-03 |
Qwen3.7 Max vs Alternatives
| Model | MMLU-Pro | SWE-Bench Verified | HLE | License |
|---|---|---|---|---|
| Qwen3.7 Max | 89.6% | 80.4% | 41.4% | Proprietary |
| Meta Muse Spark | — | 77.4% | 40.6% | Proprietary |
| Qwen3.6 27B | 86.2% | 77.2% | — | Apache 2.0 |
Qwen3.7 Max leads on SWE-Bench Verified (80.4% vs 77.4%) vs Muse Spark. For open-source alternative: Qwen3.6 27B (86.2% MMLU-Pro, 77.2% SWE-Bench Verified).
Related Models & Benchmarks
- Compare: Meta Muse Spark · Qwen3.6 27B · Qwen3.6 Plus
- Benchmarks: MMLU-Pro · SWE-Bench Verified · Humanity's Last Exam
Frequently Asked Questions
What is Qwen3.7 Max?
Alibaba's March 2026 API model scoring 89.6% MMLU-Pro, 80.4% SWE-Bench Verified, 41.4% HLE. Proprietary — strong coding + knowledge.Specs from Alibaba's Qwen3.7 Max release (March 2026) and Benchgen evaluations. Last updated 2026-07-24.