Skip to main content
Radar

Every change to the models you run, with its source and its date. Releases, price changes, retirements, API changes, and incidents.Every change to the models you run, with its source.

Follow model changes
BenchLM

Discrete Reasoning Over Paragraphs (DROP)

A reading-comprehension benchmark requiring discrete reasoning over paragraphs, reported in DeepSeek-V4 base-model evaluations.

Data verified 33 confirmed releases in the last 30 daysSee provider release alerts

Benchmark score on DROP — September 15, 2026

We mirror the published score view for DROP. DeepSeek V4 Pro Base leads the public snapshot at 88.7%, followed by DeepSeek V4 Flash Base (88.6%) and Celeris-1 (81.4%). We do not use these results to rank models overall.

1Open weights

DeepSeek V4 Pro Base

DeepSeek

88.7%
Context 1M
2Open weights

DeepSeek V4 Flash Base

DeepSeek

88.6%
Context 1M
3Proprietary

Celeris-1

Celeris

81.4%
Context 128K

4 modelsReasoningCurrentDisplay onlyUpdated September 15, 2026

Benchmark score table (4 models)

Score
1
DeepSeek V4 Pro BaseDeepSeek · Open weight
88.7%
2
DeepSeek V4 Flash BaseDeepSeek · Open weight
88.6%
3
Celeris-1Celeris · Closed
81.4%
4
Soofi S 30B-A3BSoofi Project · Open weight
66.5%

The published DROP snapshot places DeepSeek V4 Pro Base first at 88.7%. The third row is 7.3 points behind. The broader top-10 range is 22.2 points, so the table still separates the published systems.

4 models have been evaluated on DROP. The benchmark falls in the Reasoning category. This category carries a 17% weight in BenchLM.ai's overall scoring system. DROP is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About DROP

Year

2026

Tasks

Paragraph reasoning questions

Format

F1

Difficulty

Reading and numerical reasoning

BenchLM stores DROP as a display-only provider-table row when exact values are published in DeepSeek-V4 evaluations.

DeepSeek-V4 Technical Report

BenchLM freshness & provenance

Version

DROP 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 DROP measure?

A reading-comprehension benchmark requiring discrete reasoning over paragraphs, reported in DeepSeek-V4 base-model evaluations.

Which model scores highest on DROP?

DeepSeek V4 Pro Base by DeepSeek currently leads with a score of 88.7% on DROP.

How many models are evaluated on DROP?

4 AI models have been evaluated on DROP on BenchLM.

Compare Top Models on DROP

Celeris-1 vs Soofi S 30B-A3B
Last updated: September 15, 2026 · BenchLM version DROP 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.