Model Details
Qwen3 235B A22B Instruct 2507
Quick answer: Qwen3 235B A22B Instruct 2507 is the July 2025 checkpoint of Qwen3 235B A22B, scoring 83% MMLU-Pro, 79.2% Arena Hard v2, 70.9% BFCL-v3, and 41.8% ARC-AGI. Apache 2.0.
At a Glance
Where Qwen3 235B A22B Instruct 2507 leads
- 79.2% Arena Hard v2 — strong instruction quality
- 70.9% BFCL-v3 — good function calling
- 83% MMLU-Pro — solid academic knowledge
- Apache 2.0 — fully open
- July 2025 checkpoint improvements over May 2025 base
Where it lags
- 41.8% ARC-AGI — limited abstract reasoning
- 235B total params: significant hosting requirements
Best for: July 2025 improvements over the base Qwen3 235B; Arena Hard-heavy instruction pipelines; open-source BFCL function calling.
What Qwen3 235B A22B Instruct 2507 Is
Qwen3 235B A22B Instruct 2507 is the July 2025 (2507) updated checkpoint of the Qwen3 235B A22B instruct model. The "2507" suffix indicates a July 2025 re-release with instruction tuning improvements over the original May 2025 release.
With 79.2% Arena Hard v2 and 70.9% BFCL-v3, this checkpoint shows better instruction following than the base Qwen3 235B A22B release.
Specifications
| Field | Value |
|---|---|
| Organization | Alibaba |
| License | Apache 2.0 |
| HuggingFace | Qwen/Qwen3-235B-A22B-Instruct-2507 |
| Release date | July 2025 |
| Parameters | 235B total / 22B active (MoE) |
| Modality | Text only |
Pricing
Open weights under Apache 2.0 — self-host at no cost.
Public Benchmark Scores
| Benchmark | Score | Source | Date |
|---|---|---|---|
| MMLU-Pro | 83% | Benchgen evaluation | 2025-07 |
| Arena Hard v2 | 79.2% | Benchgen evaluation | 2025-07 |
| BFCL-v3 | 70.9% | Benchgen evaluation | 2025-07 |
| ARC-AGI | 41.8% | Benchgen evaluation | 2025-07 |
Qwen3 235B 2507 vs Alternatives
| Model | MMLU-Pro | Arena Hard | BFCL-v3 | License |
|---|---|---|---|---|
| Qwen3 235B A22B Instruct 2507 | 83% | 79.2% | 70.9% | Apache 2.0 |
| Qwen3 235B A22B | 68.2% | — | — | Apache 2.0 |
| Qwen3.6 Plus | — | — | — | Proprietary |
Qwen3 235B 2507 vs the base Qwen3 235B: MMLU-Pro 83% vs 68.2% — substantial July 2025 improvement. Use the 2507 checkpoint for better instruction following.
Related Models & Benchmarks
- Compare: Qwen3 235B A22B · Qwen3.6 Plus
- Benchmarks: MMLU-Pro · Arena Hard v2 · BFCL-v3
Frequently Asked Questions
What is Qwen3 235B A22B Instruct 2507?
Alibaba's July 2025 checkpoint of Qwen3 235B, scoring 83% MMLU-Pro, 79.2% Arena Hard v2, 70.9% BFCL-v3. Apache 2.0 — improved instruction tuning vs base.Specs from Alibaba's Qwen3 235B A22B Instruct 2507 release (July 2025) and Benchgen evaluations. Last updated 2026-07-24.
Benchmark Leaderboards
This model isn’t on any benchmark leaderboard yet.