DeepSeek-V4.1-FlashvsGLM-5.3
Across 12 shared benchmarks, DeepSeek-V4.1-Flash leads overall: DeepSeek-V4.1-Flash wins 11, GLM-5.3 wins 1, with 0 ties and an average score difference of +2.57.
DeepSeek-V4.1-Flash
DeepSeek-AI · 2026-09-10 · Multimodal model
GLM-5.3
智谱AI · 2026-08-14 · Reasoning model
DeepSeek-V4.1-Flash11 wins(92%)(8%)1 winGLM-5.3
Benchmark scores
Grouped by capability, sorted by largest gap within each. 12 shared benchmarks.
AI Agent - Tool Usage
DeepSeek-V4.1-Flash 3/4| Benchmark | DeepSeek-V4.1-Flash | GLM-5.3 | Diff |
|---|---|---|---|
| Terminal-Bench 4.0 | 31.207 / 16Max (With Tools) | 37.906 / 16Max (With Tools) | -6.70 |
| CyberGym | 88.101 / 8Max (With Tools) | 84.502 / 8Max (With Tools) | +3.60 |
| Terminal-Bench 2.1 | 90.601 / 53Max (With Tools) | 88.208 / 53Max (With Tools) | +2.40 |
| Terminal-Bench 3.0 | 304 / 11Max (With Tools) | 28.306 / 11Max (With Tools) | +1.70 |
Coding and Software Engineer
DeepSeek-V4.1-Flash 3/3| Benchmark | DeepSeek-V4.1-Flash | GLM-5.3 | Diff |
|---|---|---|---|
| NL2Repo-Bench | 65.402 / 16Max (With Tools) | 586 / 16Max (With Tools) | +7.40 |
| DeepSWE | 74.202 / 38Max (With Tools) | 66.9015 / 38Max (With Tools) | +7.30 |
| Program Bench | 20.308 / 11Max (With Tools) | 199 / 11Max (With Tools) | +1.30 |
Agent Capability
DeepSeek-V4.1-Flash 1/1| Benchmark | DeepSeek-V4.1-Flash | GLM-5.3 | Diff |
|---|---|---|---|
| ExploitGym (budget unspecified) | 15.302 / 4Max (With Tools) | 153 / 4Max (With Tools) | +0.30 |
Agent Level Benchmark
DeepSeek-V4.1-Flash 1/1| Benchmark | DeepSeek-V4.1-Flash | GLM-5.3 | Diff |
|---|---|---|---|
| Agents' Last Exam | 31.806 / 19Max (With Tools) | 28.508 / 19Max (With Tools) | +3.30 |
General Evaluation
DeepSeek-V4.1-Flash 1/1| Benchmark | DeepSeek-V4.1-Flash | GLM-5.3 | Diff |
|---|---|---|---|
| GPQA Diamond | 90.9039 / 274Max (No Tools) | 88.1070 / 274Max (No Tools) | +2.80 |
General Knowledge
DeepSeek-V4.1-Flash 1/1| Benchmark | DeepSeek-V4.1-Flash | GLM-5.3 | Diff |
|---|---|---|---|
| HLE | 63.903 / 197Max (With Tools) | 62.505 / 197Max (With Tools) | +1.40 |
Productivity Knowledge
DeepSeek-V4.1-Flash 1/1| Benchmark | DeepSeek-V4.1-Flash | GLM-5.3 | Diff |
|---|---|---|---|
| AutomationBench | 54.801 / 17Max (With Tools) | 48.805 / 17Max (With Tools) | +6 |
Specs
| Field | DeepSeek-V4.1-Flash | GLM-5.3 |
|---|---|---|
| Publisher | DeepSeek-AI | 智谱AI |
| Release date | 2026-09-10 | 2026-08-14 |
| Model type | Multimodal model | Reasoning model |
| Architecture | MoE | MoE |
| Parameters | 552B | 744B |
| Context length | 1M | 1M |
| Max output | 384K | 128K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | DeepSeek-V4.1-Flash | GLM-5.3 |
|---|---|---|
| Text input | ¥1 / 1M tokens | $1.4 / 1M tokens |
| Text output | ¥4 / 1M tokens | $4.4 / 1M tokens |
| Cache read | ¥0.02 / 1M tokens | $0.26 / 1M tokens |
Summary
- DeepSeek-V4.1-Flashleads in:AI Agent - Tool Usage (3/4), Coding and Software Engineer (3/3), Agent Capability (1/1), Agent Level Benchmark (1/1), General Evaluation (1/1), General Knowledge (1/1), Productivity Knowledge (1/1)
On average across the 12 shared benchmarks, DeepSeek-V4.1-Flash scores 2.57 higher.
Largest single-benchmark gap: NL2Repo-Bench — DeepSeek-V4.1-Flash 65.40 vs GLM-5.3 58 (+7.40).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.