qoder No sharer online

gm51model on Qoder.

Qoder serves gm51model at $1.40 per million input tokens and $4.40 per million output. No sharer is online for it at the moment.

Input /Mtok

$1.40

Output /Mtok

$4.40

Context

—

Sharers online

0

01Overview

Qoder serves gm51model at $1.40 per million input tokens and $4.40 per million output. No sharer is online for it at the moment.

02Pricing

Per million tokens, in USD. This is the rate you are charged.
Input$1.40
Output$4.40
Cache read$0.260

03Capabilities

Supported parametersmax_tokens, tools

Anything not listed here, we don't publish for this model yet.

05Usage

This model has not been served through Aile yet.

06Call it

Any OpenAI- or Anthropic-compatible SDK works. Use this exact model id:

curlTerminal
curl https://aile.sh/v1/chat/completions \
  -H "Authorization: Bearer $AILE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qoder/gm51model",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Quickstart · API reference

07Alternatives

Other providers serve gm51model too, at their own rates — compare every provider for gm51model.

Other models from Qoder:

ModelProviderInput /MtokOutput /MtokStatus
autoqoder——Offline
ultimateqoder——Offline
performanceqoder——Offline
efficientqoder——Offline
liteqoder——Offline
qmodelqoder$0.400$1.60Online
qmodel_latestqoder$2.50$7.50Online
qmodel_38maxqoder$2.00$6.00Online
qfmodelqoder$0.150$0.470Online
dmodelqoder$1.32$3.96Online
dfmodelqoder$0.300$1.20Online
gmodelqoder$1.40$4.40Online

Similar published rates elsewhere:

ModelProviderInput /MtokOutput /MtokStatus
deepseek/deepseek-v4.1-flashlyceum$0.500$1.50Offline
sference/deepseek-v4.1-flashrequesty$0.500$1.50Offline
tensorx/deepseek-v4.1-flashrequesty$0.500$1.50Offline
lyceum/deepseek-v4.1-flashrequesty$0.500$1.50Offline
hf:deepseek-ai/DeepSeek-V4.1-Flashsynthetic$0.600$1.20Offline
anthropic/claude-opus-5-5aimlapi$4.00$20.00Offline