openrouter-benchmarks

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

OpenRouter Benchmarks

OpenRouter 基准测试

Use OpenRouter's unified benchmarks endpoint to answer benchmark-backed model ranking and model-selection questions. The endpoint aggregates Artificial Analysis and Design Arena data and returns citation metadata that should be preserved when reporting results.

使用OpenRouter的统一基准测试端点来解答基于基准的模型排名和模型选择问题。该端点聚合了Artificial Analysis和Design Arena的数据，并返回引用元数据，报告结果时应保留这些元数据。

Prerequisites

前提条件

Set

OPENROUTER_API_KEY

to any valid OpenRouter API key. Benchmarks do not require a management key.

bash

export OPENROUTER_API_KEY=sk-or-v1-...

将

OPENROUTER_API_KEY

设置为任意有效的OpenRouter API密钥。基准测试不需要管理密钥。

bash

export OPENROUTER_API_KEY=sk-or-v1-...

Decision Tree

决策树

User wants to...	Action
See benchmark-ranked models across sources	Call `GET /api/v1/benchmarks` and preserve source/citation metadata
Choose a model for an app/use case	Check whether Artificial Analysis or Design Arena contains a relevant signal; say when no direct benchmark exists
Find best coding, intelligence, or agentic models	Use `task_type=coding` , `task_type=intelligence` , or `task_type=agentic`
Query Artificial Analysis only	Use `source=artificial-analysis`
Query Design Arena only	Use `source=design-arena` , plus `arena` and `category` when relevant
Get raw API-shaped data for integration work	Return the raw `data` / `meta` shape from the endpoint
Understand all response fields or direct curl usage	Read `references/benchmarks-api.md`

Use

openrouter-models

instead when the user needs pricing, context length, supported parameters, modalities, or provider endpoint performance without asking for benchmark rankings.

For creative writing, storytelling, or editorial apps, this endpoint currently has no direct writing-quality benchmark. Treat Artificial Analysis

intelligence_index

as a weak general-capability signal, and use

agentic_index

only if the app performs multi-step planning/revision. Do not imply that Design Arena visual/code categories measure prose quality.

用户需求	操作
查看跨来源的基准排名模型	调用 `GET /api/v1/benchmarks` 并保留来源/引用元数据
为应用/用例选择模型	检查Artificial Analysis或Design Arena是否包含相关信号；若不存在直接基准则说明情况
寻找最佳编码、智能或Agent类模型	使用 `task_type=coding` 、 `task_type=intelligence` 或 `task_type=agentic`
仅查询Artificial Analysis	使用 `source=artificial-analysis`
仅查询Design Arena	使用 `source=design-arena` ，必要时添加 `arena` 和 `category` 参数
获取用于集成工作的原始API格式数据	返回端点的原始 `data` / `meta` 结构
了解所有响应字段或直接使用curl	阅读 `references/benchmarks-api.md`

当用户需要定价、上下文长度、支持的参数、模态或供应商端点性能，且未询问基准排名时，请改用

openrouter-models

。

对于创意写作、故事创作或编辑类应用，当前该端点尚无直接的写作质量基准。可将Artificial Analysis的

intelligence_index

视为较弱的通用能力信号，仅当应用执行多步骤规划/修订时才使用

agentic_index

。请勿暗示Design Arena的视觉/代码类别可衡量散文质量。

Availability Gate

可用性检查

Do not recommend a benchmark-ranked model until it passes an availability check through the models/endpoints API. Benchmark rows can contain dated or benchmark-specific

model_permaslug

values that are useful for attribution but are not always the exact routable OpenRouter model ID.

Before recommending a benchmark candidate:

Check
```
GET /api/v1/models
```
for an exact
```
id
```
match to the benchmark
```
model_permaslug
```
.
If there is no exact
```
id
```
match but a model has
```
canonical_slug
```
equal to the benchmark
```
model_permaslug
```
, treat the benchmark row as evidence for that model family, not as a directly recommendable ID. Use the model's actual
```
id
```
only after verifying availability.

Check

GET /api/v1/models/{author}/{slug}/endpoints

or use

openrouter-models

get-endpoints.ts

for provider status.

Prefer candidates with at least one clearly usable endpoint. If all endpoints are degraded, have
```
uptime_last_30m: 0
```
, or the OpenRouter model page/API indicates the model is unavailable, exclude it from primary recommendations and explain that the benchmark result is not currently actionable.
When availability is ambiguous, say so and recommend a verified available alternative instead of presenting the benchmark leader as the default choice.

Do not rely on endpoint

status: 0

alone. Model-level availability signals such as routing error messages, warning messages, zero request limits, empty endpoint lists, or provider-specific access restrictions can make a benchmark leader non-actionable even when one endpoint appears operational. If availability signals disagree, explain the ambiguity and avoid making that model the primary recommendation.

在推荐基准排名的模型之前，需通过models/endpoints API进行可用性检查。基准测试行可能包含过时或特定于基准的

model_permaslug

值，这些值可用于归因，但并非始终是可直接路由的OpenRouter模型ID。

推荐基准候选模型前：

检查
```
GET /api/v1/models
```
是否存在与基准
```
model_permaslug
```
完全匹配的
```
id
```
。
若没有完全匹配的
```
id
```
，但某个模型的
```
canonical_slug
```
等于基准
```
model_permaslug
```
，则将基准测试行视为该模型系列的证据，而非可直接推荐的ID。仅在验证可用性后使用模型的实际
```
id
```
。

检查

GET /api/v1/models/{author}/{slug}/endpoints

或使用

openrouter-models

的

get-endpoints.ts

获取供应商状态。

优先选择至少有一个清晰可用端点的候选模型。若所有端点均降级、
```
uptime_last_30m: 0
```
，或OpenRouter模型页面/API显示模型不可用，则将其排除在主要推荐之外，并说明该基准结果当前无法实际使用。
当可用性不明确时，需说明情况并推荐已验证可用的替代方案，而非将基准排名第一的模型作为默认选择。

请勿仅依赖端点的

status: 0

。模型级别的可用性信号（如路由错误消息、警告消息、零请求限制、空端点列表或供应商特定的访问限制）可能导致基准排名第一的模型无法实际使用，即使某个端点看似正常运行。若可用性信号存在冲突，请说明这种模糊性，避免将该模型作为主要推荐。

API Usage

API 使用

Query parameters:

Flag	Values	Notes
`source`	`artificial-analysis` , `design-arena`	Omitting it returns all sources.
`task_type`	`coding` , `intelligence` , `agentic`	Maps to source-specific indices/categories.
`arena`	`models` , `builders` , `agents`	Design Arena only; defaults server-side to `models` .
`category`	`codecategories` , `uicomponent` , `gamedev` , `3d` , `dataviz` , `image` , `video` , `svg` , etc.	Design Arena only.
`max_results`	positive integer	Maximum number of rows returned by the API.

Always preserve

meta.citation

meta.source_url

, and

meta.as_of

; include attribution when republishing benchmark data.

When results include both sources, do not present them as a single absolute leaderboard: Artificial Analysis indices and Design Arena ELO use different scales. Compare within each source, or rerun with

source=artificial-analysis

source=design-arena

for a source-specific ranking.

查询参数：

标志	取值	说明
`source`	`artificial-analysis` , `design-arena`	省略时返回所有来源的数据。
`task_type`	`coding` , `intelligence` , `agentic`	与来源特定的指数/类别对应。
`arena`	`models` , `builders` , `agents`	仅适用于Design Arena；服务器端默认值为 `models` 。
`category`	`codecategories` , `uicomponent` , `gamedev` , `3d` , `dataviz` , `image` , `video` , `svg` 等	仅适用于Design Arena。
`max_results`	正整数	API返回的最大行数。

请始终保留

meta.citation

、

meta.source_url

和

meta.as_of

；重新发布基准数据时需包含归因信息。

当结果包含两个来源的数据时，请勿将其展示为单一的绝对排行榜：Artificial Analysis的指数与Design Arena的ELO使用不同的评分标准。请在每个来源内部进行比较，或重新运行查询并指定

source=artificial-analysis

或

source=design-arena

以获取特定来源的排名。

Interpreting Results

结果解读

Artificial Analysis rows include
```
intelligence_index
```
,
```
coding_index
```
, and
```
agentic_index
```
; higher is better.
Design Arena rows include
```
elo
```
,
```
win_rate
```
,
```
avg_generation_time_ms
```
,
```
arena
```
,
```
category
```
, and
```
tournament_stats
```
; higher
```
elo
```
/
```
win_rate
```
is better, lower generation time is faster.
```
pricing.prompt
```
and
```
pricing.completion
```
are USD per token as decimal strings. Multiply by 1,000,000 for per-million-token costs.
```
model_permaslug
```
identifies the benchmarked model entry. Verify it against
```
GET /api/v1/models
```
before using it as a chat/completions model ID.
```
meta.model_count
```
counts unique models in the response, which can differ from
```
data.length
```
when multiple Design Arena categories are returned.

Artificial Analysis的行包含
```
intelligence_index
```
、
```
coding_index
```
和
```
agentic_index
```
；数值越高越好。
Design Arena的行包含
```
elo
```
、
```
win_rate
```
、
```
avg_generation_time_ms
```
、
```
arena
```
、
```
category
```
和
```
tournament_stats
```
；
```
elo
```
/
```
win_rate
```
越高越好，生成时间越短速度越快。
```
pricing.prompt
```
和
```
pricing.completion
```
为每token的美元成本，以十进制字符串表示。乘以1,000,000可得到每百万token的成本。
```
model_permaslug
```
标识被基准测试的模型条目。在将其用作聊天/补全模型ID之前，请与
```
GET /api/v1/models
```
进行验证。
```
meta.model_count
```
统计响应中的唯一模型数量，当返回多个Design Arena类别时，该数值可能与
```
data.length
```
不同。

Direct API Call

直接调用API

bash

curl 'https://openrouter.ai/api/v1/benchmarks?source=artificial-analysis&task_type=coding&max_results=10' \
  -H "Authorization: Bearer $OPENROUTER_API_KEY"

Read

references/benchmarks-api.md

when implementing against the raw API or handling source-specific response shapes.

bash

curl 'https://openrouter.ai/api/v1/benchmarks?source=artificial-analysis&task_type=coding&max_results=10' \
  -H "Authorization: Bearer $OPENROUTER_API_KEY"

当基于原始API进行开发或处理特定来源的响应结构时，请阅读

references/benchmarks-api.md

。