Catalog status checked 2026-09-18

Choose a model ID, then prove it with your workload.

The catalog records route IDs and per-row effective dates. It does not rank model quality or promise that a route will accept your payload.

Start with the workload.

Use the provider directory to find an exact model ID, then compare the dated rate and run the same representative payload.

Provider family, model ID, dated rate

Provider directory and rate tables

DeepSeek

DeepSeek model rates
Model IDInputCache HitOutputEffective date
deepseek-flash$0.7$0.0233$2.12026-09-10
deepseek-v3.2$0.154Not listed$0.3082026-08-27
deepseek-v4-flash$0.638$0.0202884$1.9142026-08-27
deepseek-v4-pro$1.914$0.0637362$5.7422026-08-27

GLM

GLM model rates
Model IDInputCache HitOutputEffective date
glm-4.5$0.6975$0.1800003$2.172026-08-27
glm-4.5-air$0.6525$0.1620001$2.032026-08-27
glm-4.6$0.93$0.2199999$3.412026-08-27
glm-4.7$0.93$0.2199999$3.412026-08-27
glm-5$1.55$0.4000008$4.962026-08-27
glm-5-turbo$1.8$0.4800006$5.42026-08-27
glm-5.1$2.1$0.680001$6.62026-08-27

Kimi

Kimi model rates
Model IDInputCache HitOutputEffective date
kimi-k2.6$1.09$0.1868571$4.59982026-08-27
kimi-k2.7-code$1.89$0.2850007$62026-08-27
kimi-k2.7-code-highspeed$4.725$0.7125016$152026-08-27
kimi-k3$4.5$0.9$22.52026-08-27

ERNIE

ERNIE model rates
Model IDInputCache HitOutputEffective date
ernie-4.5-turbo$0.19726$0.049315$0.74520882026-08-27
ernie-5.0$2.46575$0.2465754$9.315132026-08-27

MiniMax

MiniMax model rates
Model IDInputCache HitOutputEffective date
MiniMax-M2$0.46866$0.046866$1.874642026-08-27
MiniMax-M2.1$0.46866$0.046866$1.874642026-08-27
MiniMax-M2.1-highspeed$0.93732$0.046866$3.749282026-08-27
MiniMax-M2.5$0.46866$0.046866$1.874642026-08-27
MiniMax-M2.5-highspeed$0.93732$0.046866$3.749282026-08-27
MiniMax-M2.7$0.45304$0.090608$1.812162026-08-27
MiniMax-M2.7-highspeed$0.93732$0.093732$3.749282026-08-27
MiniMax-M3$0.90608$0.181216$3.624322026-08-27

Qwen

Qwen model rates
Model IDInputCache HitOutputEffective date
qwen-72b-chat$4.46342Not listed$4.463422026-08-27
qwen-audio-3.0-realtime-flash$0.6695133Not listed$0.66951332026-08-27
qwen-audio-3.0-realtime-plus$1.11586Not listed$1.115862026-08-27
qwen-image-2.0-pro-2026-04-22$0.1227441Not listed$0.12274412026-08-27
qwen-image-3.0$0.0401708Not listed$0.04017082026-08-27
qwen-image-3.0-pro$0.0557928Not listed$0.05579282026-08-27
qwen1.5-72b-chat$4.46342Not listed$4.463422026-08-27
qwen3-max$1.5622Not listed$6.248792026-08-27
qwen3.5-122b-a10b$0.49315Not listed$3.726042026-08-27
qwen3.5-27b$0.443836Not listed$3.353452026-08-27
qwen3.5-omni-flash$0.4909764Not listed$2.968182026-08-27
qwen3.5-plus$0.4463422Not listed$2.678052026-08-27
qwen3.6-27b$0.6695133Not listed$4.017082026-08-27
qwen3.6-35b-a3b$0.401708Not listed$2.410252026-08-27
qwen3.6-flash$0.2678053Not listed$1.606832026-08-27
qwen3.7-flash-2026-07-15$0.2678053Not listed$1.071222026-08-27
qwen3.7-max-2026-05-20$2.67805Not listed$8.034162026-08-27
qwen3.7-max-2026-06-08$2.67805Not listed$8.034162026-08-27
qwen3.7-plus$0.4463422Not listed$1.785372026-08-27
qwen3.7-text-embedding$0.1115856Not listed$0.11158562026-08-27
qwen3.8-2.4t-a95b$2.67805Not listed$8.034162026-08-27
qwen3.8-27b$0.6695133Not listed$2.678052026-08-27
qwen3.8-max$2.67805Not listed$8.034162026-08-27

Doubao

Doubao model rates
Model IDInputCache HitOutputEffective date
doubao-seed-2-0-lite-260428$0.401708Not listed$2.410252026-08-27
doubao-seed-2-0-mini-260428$0.1785369Not listed$1.785372026-08-27
doubao-seed-2-1-pro-260628$1.33903Not listed$6.695132026-08-27
doubao-seed-evolving$1.33903Not listed$6.695132026-08-27

StepFun

StepFun model rates
Model IDInputCache HitOutputEffective date
step-3.5-flash$0.21$0.042$0.632026-08-27
step-3.5-flash-2603$0.21$0.042$0.632026-08-27
step-3.7-flash$0.4$0.08$2.42026-08-27

MiMo

MiMo model rates
Model IDInputCache HitOutputEffective date
xiaomi/mimo-v2.5-pro$1.5622Not listed$4.686592026-08-27
Read each row as a reference.
Text routes show USD per 1M token units with an effective date for each row. Image and realtime audio routes can use route-specific billable units. Confirm the live rate, effective group, and workload behavior before production use.