Skip to main content

Benchmark profile

IDE-Bench

An 80-task software-engineering benchmark across eight repositories that tests whether autonomous IDE agents can explore, edit, run, and verify code changes end to end.

How we show IDE-Bench

We mirror AfterQuery's IDE-Bench pass@1 table captured on July 18, 2026 snapshot. The benchmark contains 80 software-engineering tasks across eight repositories, and the snapshot preserves the published score margins for all 15 rows.

IDE-Bench measures model performance inside an autonomous IDE-agent harness. We keep the table display-only because the result belongs to the complete model, tools, container, and grading setup rather than to a normalized base-model run.

15 model rows80 tasks8 repositoriesPass@1 with marginsDisplay only

Pass@1 accuracy on IDE-Bench — July 18, 2026 snapshot

BenchLM mirrors the published pass@1 accuracy view for IDE-Bench. Claude Sonnet 4.5 leads the public snapshot at 87.50% , followed by GPT-5.2 (85.00%) and Claude Opus 4.5 (83.75%). BenchLM does not use these results to rank models overall.

15 modelsCodingCurrentDisplay onlyUpdated July 18, 2026 snapshot

Pass@1 accuracy table (15 models)

Score
1
Claude Sonnet 4.5Anthropic · Closed
87.50%
2
GPT-5.2OpenAI · Closed
85.00%
3
Claude Opus 4.5Anthropic · Closed
83.75%
4
Claude Haiku 4.5Anthropic · Closed
78.75%
5
GPT-5.1 Codex MaxOpenAI · Closed
73.75%
6
Qwen3 MaxAlibaba · Closed
65.00%
7
57.50%
8
Gemini 3 ProGoogle · Closed
55.00%
9
Grok 4.1 FastxAI · Closed
35.00%
10
DeepSeek-V3.2DeepSeek · Open weight
31.25%
11
DeepSeek-R1DeepSeek · Open weight
20.00%
12
Grok Code Fast 1xAI · Closed
11.25%
13
Llama 4 MaverickMeta · Open weight
2.50%
14
Llama 4 ScoutMeta · Open weight
2.50%
15
0.00%

The published IDE-Bench snapshot places Claude Sonnet 4.5 first at 87.50%. The third row is 3.75 points behind. The broader top-10 range is 56.25 points, so the table still separates the published systems.

15 models have been evaluated on IDE-Bench. The benchmark falls in the Coding category. This category carries a 20% weight in BenchLM.ai's overall scoring system. IDE-Bench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About IDE-Bench

Year

2026

Tasks

80 tasks across 8 repositories

Format

Autonomous IDE-agent task completion (pass@1)

Difficulty

End-to-end software engineering

The published AfterQuery table reports pass@1 accuracy with a ± margin for 15 models. We preserve the source rows as display-only agent results because the model operates through an IDE tool harness inside a container rather than through a normalized model-only prompt.

BenchLM freshness & provenance

Version

IDE-Bench 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does IDE-Bench measure?

An 80-task software-engineering benchmark across eight repositories that tests whether autonomous IDE agents can explore, edit, run, and verify code changes end to end.

Which model leads the published IDE-Bench snapshot?

Claude Sonnet 4.5 currently leads the published IDE-Bench snapshot with 87.50% pass@1 accuracy. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on IDE-Bench?

15 AI models are included in BenchLM's mirrored IDE-Bench snapshot, based on the public leaderboard captured on July 18, 2026 snapshot.

Last updated: July 18, 2026 snapshot · mirrored from the public benchmark leaderboard

The AI models change fast. We track them for you.

A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.