# Qwen3-2507

Qwen3-2507 is an AI model released by Qwen on Jul 21 2025. It has 4B/30B/235B parameters, a 256k context window and open weight. Tracked results include 81.1% on GPQA Diamond, 74.1% on LiveCodeBench and 18.2% on Humanity's Last Exam (text only, no tools).

## Facts

| Field | Value |
| --- | --- |
| Model | Qwen3-2507 |
| Developer | Qwen |
| Release date | Monday, Jul 21 2025 |
| Licensing | Open Weight |
| Parameters | 4B/30B/235B |
| Context window | 256k |

## Tracked benchmark scores

| Benchmark | Score | Source | What it measures |
| --- | --- | --- | --- |
| LiveCodeBench | 74.1% | Lab | Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better. |
| IFEval | 87.8% | Lab | following verifiable instructions in generated responses |
| Humanity's Last Exam (text only, no tools) | 18.2% | Lab | accuracy on the 2,158 text-only questions without tools |
| AIME 2025 | 92.3% | Lab | single-answer accuracy on the 2025 invitational mathematics exam without tools |
| GPQA Diamond | 81.1% | Lab | Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. |
| MMLU-Pro | 84.4% | Lab | accuracy on the ten-option MMLU-Pro benchmark |

## About Qwen3-2507

Qwen3-2507 began on July 21, 2025 with a post-trained 235B Instruct model. Its companion 235B Thinking checkpoint followed on July 25, and smaller 30B and 4B sizes joined the same refresh by August 6.

This release separated direct responses from extended reasoning, which the April Qwen3 models had supported in the same checkpoint. The 235B Thinking model scored 92.3 on AIME 2025 and 74.1 on LiveCodeBench v6 in Qwen's model card. Qwen later documented a context extension to roughly one million tokens for the larger 2507 checkpoints.

## Questions and answers

### When was Qwen3-2507 released?

Qwen3-2507 was released by Qwen on Monday, Jul 21 2025.

### Who made Qwen3-2507?

Qwen3-2507 was built by Qwen. Alibaba's AI lab, building the Qwen family. The most prolific publisher of open-weight models of any major lab, alongside a closed Max and Plus tier sold through Alibaba Cloud.

### What benchmark scores did Qwen3-2507 get?

Qwen3-2507 reports 6 tracked benchmark scores — LiveCodeBench: 74.1%; IFEval: 87.8%; Humanity's Last Exam (text only, no tools): 18.2%; AIME 2025: 92.3%; GPQA Diamond: 81.1%; MMLU-Pro: 84.4%. Tracked scores may come from lab reports or independent benchmarks; source details accompany the benchmark data.

### What is the context window of Qwen3-2507?

Qwen3-2507 has a context window of 256k. Providers may specify separate input and output token limits.

### How many parameters does Qwen3-2507 have?

Qwen3-2507 is reported at 4B/30B/235B parameters.

### Is Qwen3-2507 open source?

Partly. Qwen3-2507 is an open-weight model: the trained weights are free to download and run locally or on your own infrastructure, but the training data and code are not fully released and the license may restrict some commercial uses. It is not open source in the strict sense.

### What came before and after Qwen3-2507?

Qwen's previous tracked release was Qwen3 on Apr 29 2025, 83 days earlier. It was followed by Qwen3-Coder on Jul 22 2025.


---

Canonical page: https://aireleasetracker.com/model/qwen/qwen3-2507
Site index: https://aireleasetracker.com/llms.txt
Source: AI Release Tracker (https://aireleasetracker.com). Most benchmark scores come from lab launch material; gathered results identify the leaderboard that published them.
