CO

Composer 2.5

Coding model

Composer 2.5

Release date: 2026-05-18Updated: 2026-05-19 16:00:13730
Live demoGitHubHugging FaceCompare
Parameters
Not disclosed
Context length
200K
Chinese support
Not supported
Reasoning ability

Composer 2.5 is an AI model published by Cursor, released on 2026-05-18, for Coding model, and 200K context length, under the 不开源 license, with a 79.80 score on SWE-bench Multilingual.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Composer 2.5

Model basics

Reasoning traces
Not supported
Thinking modes
Thinking Mode (Default)Standard Mode
Context length
200K tokens
Max output length
No data
Model type
Coding model
Modality (in / out)
No data
Release date
2026-05-18
Model file size
No data
MoE architecture
No
Total params / Active params
No data / N/A
Knowledge cutoff
No data
Composer 2.5

Open source & experience

Code license
不开源
Weights license
不开源
GitHub repo
GitHub link unavailable
Hugging Face
Hugging Face link unavailable
Live demo
No live demo
Composer 2.5

Official resources

DataLearnerAI blog
No blog post yet
Composer 2.5

API details

API speed
3/5
No public API pricing yet.
Composer 2.5

Benchmark Results

Composer 2.5 currently shows benchmark results led by SWE-bench Multilingual (2 / 20, score 79.80), Terminal Bench 2.0 (7 / 46, score 69.30). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking

AI Agent - Tool Usage

1 evaluations
Benchmark / mode
Score
Rank/total
Terminal Bench 2.0
Thinking Mode
69.30
7 / 46

Coding and Software Engineer

1 evaluations
Benchmark / mode
Score
Rank/total
79.80
2 / 20

Compare with other models

No curated comparisons for this model yet.

Want a custom combination? Open the compare tool

Composer 2.5

Publisher

Composer 2.5

Model Overview

Composer 2.5 is an AI model published by Cursor, released on 2026-05-18, for Coding model, and 200K context length, under the 不开源 license, with a 79.80 score on SWE-bench Multilingual.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code