Skip to content
AuraMatrix
← All models

Measured

What the APIs did when we called them.

A price says what a token costs. It does not say whether the API answers, how long you wait, or whether the bill matches the price list. Every hour we send each model one small request on every channel we hold a key for, and write down what happened.

Last 7 daysmeasuring since 2026-11-11 13:47 UTCupdated 2026-11-15 22:22 UTC

Calls answered4028 / 403199.9%
Charges matching the recorded list price1547 / 2330requests where the provider reported the charge
Models and channels measured41one small request each, every hour

By model and channel

ModelChannelAnsweredFirst tokenmedian / 9 in 10 withinSpeedtokens per secondCharged vs list price
Claude Fable 5.1Claude API100%106 / 1060.64 s1.8 s125—
Claude Fable 5.1OpenRouterGoogle, Azure100%107 / 1071.9 s2.5 s135as listed
Claude Fable 5.2Claude API100%106 / 1060.64 s0.84 s163—
Claude Haiku 5.5Claude API100%106 / 1060.53 s0.88 s379—
Claude Haiku 5.5OpenRouterClaude Platform on AWS, Google100%107 / 1070.56 s0.84 s376as listed
Claude Opus 5.5Claude API100%106 / 1060.69 s1.2 s161—
Claude Opus 5.5OpenRouterClaude Platform on AWS, Amazon Bedrock100%107 / 1071.3 s1.7 s162as listed
Claude Sonnet 5.5Claude API100%106 / 1060.62 s0.83 s228—
Claude Sonnet 5.5OpenRouterClaude Platform on AWS, Amazon Bedrock100%107 / 1070.57 s0.68 s242as listed
DeepSeek V4 ProDeepSeek API100%64 / 641.4 s1.7 s474—
DeepSeek V4 ProOpenRouterRelace, Baidu100%107 / 1071.8 s3.7 s499−13%
DeepSeek V4.1 FlashDeepSeek API100%64 / 641.0 s1.2 s——
DeepSeek V4.1 FlashOpenRouterWafer, InferenceNet100%107 / 1070.55 s1.1 s282−48%
Gemini 3.5 Flash-LiteGemini Developer API100%106 / 1060.65 s0.92 s570—
Gemini 3.5 Flash-LiteOpenRouterGoogle100%107 / 1070.60 s0.76 s495as listed
Gemini 3.8 FlashGemini Developer API100%106 / 1062.6 s3.5 s282—
Gemini 3.8 FlashOpenRouterGoogle, Google AI Studio100%107 / 1072.6 s4.0 s301−9.1%
Gemini 4 ArgonGemini Developer API99.1%105 / 1064.2 s9.5 s125—
Gemini 4 ArgonOpenRouterGoogle, Google AI Studio100%107 / 1073.1 s8.1 s87as listed
GLM-5.3OpenRouterWafer, Decart100%107 / 1070.71 s1.5 s253as listed
GLM-5.3Z.AI API100%64 / 641.8 s2.7 s282—
GLM-5.3-FlashOpenRouterInferenceNet, Wafer100%107 / 1070.88 s1.9 s78+276%
GLM-5.3-FlashZ.AI API100%64 / 641.6 s2.4 s168—
GPT-6 AstraOpenAI API100%106 / 1062.2 s3.1 s152—
GPT-6 AstraOpenRouterOpenAI100%107 / 1071.5 s2.0 s148as listed
GPT-6 LunaOpenAI API100%106 / 1060.95 s1.5 s166—
GPT-6 LunaOpenRouterOpenAI100%107 / 1071.2 s2.1 s225as listed
GPT-6.1 SolOpenAI API100%106 / 1061.8 s2.2 s175—
GPT-6.1 SolOpenRouterOpenAI100%107 / 1071.1 s1.5 s160as listed
GPT-6.1 TerraOpenAI API100%106 / 1061.1 s1.5 s91—
GPT-6.1 TerraOpenRouterOpenAI100%107 / 1070.95 s1.4 s95as listed
Kimi K3Kimi API100%64 / 643.7 s6.8 s166—
Kimi K3.1 ProKimi API98.4%63 / 642.8 s3.7 s88—
Kimi K3.1 ProOpenRouterWafer, InferenceNet100%107 / 1071.1 s2.5 s285−6.5%
Llama 4 MaverickOpenRouterParasail99.1%106 / 1070.34 s0.49 s90as listed
MiniMax M3MiniMax API100%64 / 640.65 s1.0 s164—
MiniMax M3OpenRouterCoreWeave, Parasail100%107 / 1070.72 s2.2 s200−21%
MiniMax M3.1 FlashMiniMax API100%64 / 641.2 s3.0 s155—
MiniMax M3.1 FlashOpenRouterTogether, Nebius100%107 / 1070.55 s0.97 s666+58%
Qwen3.8 FlashOpenRouterAlibaba100%107 / 1071.3 s2.1 s170as listed
Qwen3.8 MaxOpenRouterAlibaba100%107 / 1072.5 s3.4 s234as listed

How to read it

Answered
The share of our hourly calls that returned an answer. Failed calls count against this; calls refused because of our own account (no credit, or a key without access) are excluded.
First token
Time from sending the request to the first visible text, measured from our Los Angeles server. Models that think before answering take longer by design.
Speed
Tokens per second while writing, from one longer request a day. A dash means no speed run has been made yet.
Charged vs list price
Where the provider reports the charge for a request, we compare that amount with the list price we had on record that day. A marketplace routes to different providers at different prices, so its charge can be above or below its listed price. A dash means the provider does not report per-request charges.
What this is not
This does not test answer quality or performance under load: one small request from one location says nothing about behaviour under heavy use or from other regions.