dsh-llm-latency
Per-vendor and per-model LLM latency telemetry and benchmark dashboard for DeepSeek Harness.
Install
$ dsh plugin --profile web add github:shengbinxu/dsh-llm-latencyPlugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Passive telemetry capturing TTFT, end-to-end stream duration, tokens per second, and prompt cache hit rates
- Failure classification across rate limits (429), timeouts, 5xx server errors, and user aborts
- Cross-vendor time-window comparisons with P50/P90/P95/P99 latency percentiles and median confidence intervals
- Side-by-side session latency and performance comparisons for single-model runs
- Local interactive HTML dashboard hosted on DeepSeek Harness web profile
- Built-in latency_report model tool for programmatic querying of latency statistics
- Searchable and filterable append-only request logging
Useful For
- Comparing real-world latency and cache performance across different model providers for the same model
- Diagnosing provider rate limits, timeouts, and upstream server errors
- Evaluating prompt caching efficiency and cost-saving metrics across vendors
- Monitoring inference bottlenecks during interactive AI agent sessions
Who It Fits
- DeepSeek Harness developers and administrators
- AI engineers optimizing LLM routing and vendor latency
- Users managing multiple model providers seeking empirical performance benchmarks
Documented Limitations
- Requires the llm service from dsh-base to intercept stream requests
- Session comparisons require single-model sessions that did not switch models during execution
- Median statistical confidence intervals require a minimum sample threshold (configurable via minSamplesForComparison)
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
package.json declares 0 runtime, 3 development, 0 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub did not report an SPDX license for this repository snapshot.
A root package.json was captured and can be inspected by the registry.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- shengbinxu/dsh-llm-latency
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 535a3d0 · 2026-08-31
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 0GitHub stars
- Forks
- 0
- Open issues
- 0
- Last commit
- 2026-08-29
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Vision routing and pixel-level visual tool suite for DeepSeek Harness agents with built-in free fallbacks.
View plugin →Browser automation plugin enabling DeepSeek Harness agents to interact with logged-in browser sessions safely via dedicated Agent Windows.
View plugin →DSH undo and crash-recovery plugin with snapshots, rollback, Safe Mode, and offline Windows tools.
View plugin →DeepSeek Harness plugin integrating ego-lite Chromium browser with 32 structured automation tools and live interactive screencast panel.
View plugin →