ai

[!NOTE|label:references:]
Introduction to Generative AI

chat tool

[!NOTE|label:references:]
如何在 Chatbox 中接入 SiliconCloud - 超简单完整教程

chatgpt

model

[!NOTE|label:references:]
token limits
models

ABBREVIATION

DESCRIPTION

RPM

Requests Per Minute

TPM

Tokens Per Minute

IPM

Images Per Minute

AMPM

Audio Megabytes Per Minute

RPD

Requests Per Day

TPD

Tokens Per Day

Batch Tokens Per Day

MODEL NAME

MAX CONTEXT

MAX OUTPUT TOKENS

gpt-3.5-turbo

16,385

4,096

gpt-3.5-turbo-16k

16,385

gpt-4

8,192

gpt-4-32k

32,768

gpt-4-turbo

128,000

4,096

gpt-4-32k

32,768

gpt-4-32k-0613

32,768

gpt-4 turbo

128,000

4,096

gpt-4o

128,000

16,384

gpt-4o-mini

128,000

100,000

gpt-4.1

1,047,576

32,768

gpt-o3

200,000

100,000

gpt-4.1 nano

1,047,576

32,768

MODEL

TOKEN LIMITS

REQUEST AND OTHER LIMITS

BATCH QUEUE LIMITS

gpt-3.5-turbo

200,000 TPM

500 RPM 10,000 RPD

2,000,000 TPD

gpt-3.5-turbo-0125

200,000 TPM

500 RPM 10,000 RPD

2,000,000 TPD

gpt-3.5-turbo-1106

200,000 TPM

500 RPM 10,000 RPD

2,000,000 TPD

gpt-3.5-turbo-16k

200,000 TPM

500 RPM 10,000 RPD

2,000,000 TPD

gpt-3.5-turbo-instruct

90,000 TPM

3,500 RPM

200,000 TPD

gpt-3.5-turbo-instruct-0914

90,000 TPM

3,500 RPM

200,000 TPD

gpt-4

10,000 TPM

500 RPM 10,000 RPD

100,000 TPD

gpt-4-0613

10,000 TPM

500 RPM 10,000 RPD

100,000 TPD

gpt-4-turbo ↪︎ gpt-4-turbo-2024-04-09 ↪︎ gpt-4-turbo-preview ↪︎ gpt-4-0125-preview ↪︎ gpt-4-1106-preview

30,000 TPM

500 RPM

90,000 TPD

gpt-4.1 ↪︎ gpt-4.1-2025-04-14

30,000 TPM

500 RPM

900,000 TPD

gpt-4.1 (long context)

200,000 TPM

100 RPM

2,000,000 TPD

gpt-4.1-mini ↪︎ gpt-4.1-mini-2025-04-14

200,000 TPM

500 RPM

2,000,000 TPD

gpt-4.1-mini (long context)

400,000 TPM

200 RPM

4,000,000 TPD

gpt-4.1-nano ↪︎ gpt-4.1-nano-2025-04-14

200,000 TPM

500 RPM

2,000,000 TPD

gpt-4.1-nano (long context)

400,000 TPM

200 RPM

4,000,000 TPD

gpt-4.5-preview ↪︎ gpt-4.5-preview-2025-02-27

125,000 TPM

1,000 RPM

50,000 TPD

gpt-4o ↪︎ gpt-4o-2024-05-13 ↪︎ gpt-4o-2024-08-06 ↪︎ gpt-4o-2024-11-20 ↪︎ gpt-4o-audio-preview ↪︎ gpt-4o-audio-preview-2024-10-01 ↪︎ gpt-4o-audio-preview-2024-12-17

30,000 TPM

500 RPM

90,000 TPD

gpt-4o-mini ↪︎ gpt-4o-mini-2024-07-18 ↪︎ gpt-4o-mini-audio-preview ↪︎ gpt-4o-mini-audio-preview-2024-12-17

200,000 TPM

500 RPM 10,000 RPD

2,000,000 TPD

API

[!NOTE|label:references:]
Data controls in the OpenAI platform
Models
Error codes

api type

[!TIP|label:curl commands:]

$ curl -s '${scheme}://${authority}${path}' \
       -H 'Authorization: Bearer ${Authorization}' \
       -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64)' \
       -H 'Origin: https://platform.openai.com' \
       -H 'Referer: https://platform.openai.com/settings/organization/limits' \
       -H 'Accept: application/json' \
       -H 'Content-Type: application/json'

TYPE

EXAMPLE

PUBLICLY DOCUMENTED

STABLE

RECOMMENDED FOR AUTOMATION

Public API

/v1/chat/completions

✅ Yes

✅ Stable

✅ Yes

Private API

/dashboard/rate_limits

❌ No

⚠️ May change

⚠️ For personal/internal use only

Internal API

/api/internal/user/settings

❌ No

❌ Unstable

❌ Not recommended

Browser-bound API

Requires session + Origin + Referer headers

❌ No

⚠️ Risky

❌ Manual use only

get authorization key

visit https://platform.openai.com/settings/organization/limits
⌥ + ⌘ + i to open the devtools
open the tab Network → Fetch/XHR, and search with keyword limit
check the item one by one, untile find the one with Authorization: Bearer sess-**************************************** header in the Request Headers
double confirm the Response tab has the model limits data

as long as found the correct item, check the Request Headers tab, it will be ( example ):

KEY

VALUE

:authority

api.openai.com

:path

/dashboard/rate_limits

:method

GET

:scheme

https

Authorization

Bearer sess-xxxxxx

[!TIP|label:other APIs]

AUTHORITY

Endpoint

SCHEME

METHOD

api.openai.com

/dashboard/billing/subscription

https

GET

api.openai.com

/dashboard/rate_limits

https

GET

api.openai.com

/dashboard/user/api_keys

https

GET

api.openai.com

/dashboard/feature_flags

https

GET

api.openai.com

/v1/dashboard/organizations/<ORGANIZATION_ID>/projects/<PROJECT_ID>/users

https

GET

curl API call example:

$ curl -s '${scheme}://${authority}${path}' \
       -H 'Authorization: Bearer ${Authorization}' \
       -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64)' \
       -H 'Origin: https://platform.openai.com' \
       -H 'Referer: https://platform.openai.com/settings/organization/limits' \
       -H 'Accept: application/json' \
       -H 'Content-Type: application/json'

# i.e.:
$ curl -s 'https://api.openai.com/dashboard/rate_limits' \
       -H 'Authorization: Bearer sess-xxxxxx' \
       -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64)' \
       -H 'Origin: https://platform.openai.com' \
       -H 'Referer: https://platform.openai.com/settings/organization/limits' \
       -H 'Accept: application/json' \
       -H 'Content-Type: application/json'

model

$ curl -s https://api.openai.com/v1/models \
       -H "Authorization: Bearer $OPENAI_API_KEY"

# or show in table format
$ curl -s https://api.openai.com/v1/models \
       -H "Authorization: Bearer $OPENAI_API_KEY" |
  jq -r '["CREATED AT", "MODEL ID"],
         (.data | sort_by(.created) | reverse[] | [(.created | strftime("%Y-%m-%d %H:%M:%S")), .id])
         | @tsv' |
  column -t -s$'\t'
CREATED AT           MODEL ID
2025-06-02 23:54:58  gpt-4o-audio-preview-2025-06-03
2025-06-02 23:43:58  gpt-4o-realtime-preview-2025-06-03
2025-05-08 03:00:57  codex-mini-latest
2025-04-24 17:50:30  gpt-image-1
...

model_limits

$ curl -s https://api.openai.com/v1/fine_tuning/model_limits \
       -H "Authorization: Bearer ${OPENAI_API_KEY}"

# with table format
$ curl -s https://api.openai.com/v1/fine_tuning/model_limits \
       -H "Authorization: Bearer $OPENAI_API_KEY" |
  jq -r '["ID", "PENDING JOBS", "MAX PENDING JOBS", "JOBS", "MAX JOBS/DAY"],
         (.data[] | [.id, .curr_pending_jobs, .max_pending_jobs, .jobs_so_far_today, .max_jobs_per_day])
         | @tsv' |
  column -t -s$'\t'
ID                       PENDING JOBS  MAX PENDING JOBS  JOBS  MAX JOBS/DAY
gpt-3.5-turbo-0125       0             3                 0     48
gpt-3.5-turbo-1106       0             3                 0     48
gpt-4.1-2025-04-14       0             3                 0     8
gpt-4.1-mini-2025-04-14  0             6                 0     24
gpt-4.1-nano-2025-04-14  0             6                 0     24
gpt-4o-2024-08-06        0             3                 0     8
gpt-4o-mini-2024-07-18   0             5                 0     24
o4-mini-2025-04-16       0             2                 0     4

# or
$ (
    echo "model | curr pending | max pending | today jobs | max daily jobs" | tr '[:lower:]' '[:upper:]';
    curl -s https://api.openai.com/v1/fine_tuning/model_limits -H "Authorization: Bearer ${OPENAI_API_KEY}" |
      jq -r '.data[] | [.id, .curr_pending_jobs, .max_pending_jobs, .jobs_so_far_today, .max_jobs_per_day] | @tsv' |
      while IFS=$'\t' read -r model curr max today maxday; do
        echo "${model} | ${curr} | ${max} | ${today} | ${maxday}";
      done
  ) | column -t -s '|'

rate_limits

[!TIP|label:references:]
rate limits in headers

$ curl -s https://api.openai.com/dashboard/rate_limits \
       -H 'Authorization: Bearer sess-s**************************************P' \
       -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64)' \
       -H 'Accept: application/json' \
       -H 'Content-Type: application/json' |
  jq -r 'to_entries |
         (["MODEL", "RPM", "TPM", "IMAGES/MIN", "AUDIO_MB/MIN", "RPD", "BATCH_TOKENS"]),
         (.[] | [
           .key,
           (.value.max_requests_per_1_minute        // "-" | if . == "-" then . else (. | tostring + " RPM") end),
           (.value.max_tokens_per_1_minute          // "-" | if . == "-" then . else (. | tostring + " TPM") end),
           (.value.max_images_per_1_minute          // "-" | if . == "-" then . else (. | tostring + " IPM") end),
           (.value.max_audio_megabytes_per_1_minute // "-" | if . == "-" then . else (. | tostring + " MB/min") end),
           (.value.max_requests_per_1_day           // "-" | if . == "-" then . else (. | tostring + " req/day") end),
           (.value.batch_1_day_max_input_tokens     // "-" | if . == "-" then . else (. | tostring + " tokens") end)
         ]) | @tsv' |
  column -t -s $'\t'

Administration API

admin key

[!TIP|label:OPENAI_ADMIN_KEY:]
openai admin key
Admin API Keys
List all organization and project API keys

# get user id, can be used for `-H "OpenAI-User: user-xxxxxxxxxxxxxxxx"`
$ curl -s https://api.openai.com/v1/organization/users \
       -H "Authorization: Bearer $OPENAI_ADMIN_KEY" \
       -H "Content-Type: application/json"

# get details of admin key
$ curl -s https://api.openai.com/v1/organization/admin_api_keys \
       -H "Authorization: Bearer ${OPENAI_ADMIN_KEY}" \
       -H "Content-Type: application/json"

organization id

# get organization id, can be used for `-H "OpenAI-Organization: org-xxxxxxxxxxxxxxxx"`
$ curl -s https://api.openai.com/v1/organizations \
       -H "Authorization: Bearer sess-s**************************************P" \
       -H "Content-Type: application/json" |
  jq -r '.data[].id'
org-G**********************o

# or with ADMIN_KEY
$ curl -s https://api.openai.com/v1/organizations \
       -H "Authorization: Bearer ${OPENAI_ADMIN_KEY}" \
       -H "Content-Type: application/json" |
  jq -r '.data[].id'
org-GUjEKO81ydwmSGORg4boXRso

group id

[!TIP|label:group id:]
to get group id, can be used for -H "OpenAI-Group: group-xxxxxxxxxxxxxxxx"

# or show in table format
$ curl -s https://api.openai.com/v1/organization/projects \
       -H "Authorization: Bearer ${OPENAI_ADMIN_KEY}" \
       -H "Content-Type: application/json" |
  jq -r '["NAME", "GROUP ID"], (.data[] | [.name, .id]) | @tsv' |
  column -t -s$'\t'
NAME             GROUP ID
Default project  proj_f**********************4
marslo           proj_E**********************m

# or via organization API
$ curl -s https://api.openai.com/v1/organizations \
       -H "Authorization: Bearer ${OPENAI_ADMIN_KEY}" \
       -H "Content-Type: application/json" |
  jq -r '["NAME", "GROUP ID"], (.data[].projects.data[] | [.title, .id]) | @tsv' |
  column -t -s$'\t'
NAME             GROUP ID
Default project  proj_f**********************4
marslo           proj_E**********************m

$ curl -s https://api.openai.com/v1/organization/projects \
       -H "Authorization: Bearer ${OPENAI_ADMIN_KEY}" \
       -H "Content-Type: application/json" |
  jq -r '.data[] | select(.name == "marslo") | .id'

# or show name and id
$ curl -s https://api.openai.com/v1/organization/projects \
       -H "Authorization: Bearer ${OPENAI_ADMIN_KEY}" \
       -H "Content-Type: application/json" |
  jq -r '.data[] | "\(.name)|\(.id)"' | column -t -s'|'

using chatgpt to generate git commits

#!/usr/bin/env bash

set -euo pipefail

model="${COMMITX_MODEL:-gpt-4-1106-preview}"
temperature="${COMMITX_TEMPERATURE:-0.3}"
OPENAI_API_KEY="${OPENAI_API_KEY:? OPENAI_API_KEY not set}"
diff=$(git --no-pager diff --color=never)
prompt="Generate a Git commit message in the Conventional Commits format, with the following structure:

<type>(<scope1,scope2,...>): short summary

- detail 1
- detail 2
- ...

Use multiple scopes if the diff includes changes in multiple areas.
Use markdown-style bullet points for details.
Do not include explanations or code.
Base it on the following diff:

$diff"

declare payload
payload=$(jq -n \
  --arg model "$model" \
  --arg temp "$temperature" \
  --arg prompt "$prompt" \
  '{
    model: $model,
    temperature: ($temp | tonumber),
    messages: [
      { role: "system", content: "You are an assistant that writes Git commit messages." },
      { role: "user", content: $prompt }
    ]
  }')


curl -s https://api.openai.com/v1/chat/completions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d "$payload" |
jq -r '.choices[0].message.content'

# vim:tabstop=2:softtabstop=2:shiftwidth=2:expandtab:filetype=sh

using chatgpt to review git diff

#!/usr/bin/env bash

set -euo pipefail

MODEL="${COMMITX_MODEL:-gpt-4-turbo}"
TEMPERATURE="${COMMITX_TEMPERATURE:-0.3}"
OPENAI_API_KEY="${OPENAI_API_KEY:? OPENAI_API_KEY not set}"
diff=$(git --no-pager diff --color=never)
prompt="You are a Principal Software Architect reviewing a Git diff.

You may not have the full code context, but please still provide a professional code review.

Focus on:
- What the changes are doing
- Code quality, style, naming, structure
- Potential bugs or anti-patterns
- Suggestions for improvement

Use markdown. Be helpful, concise, and insightful.

--- START DIFF ---
${diff}
--- END DIFF ---"

curl -s https://api.openai.com/v1/chat/completions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d "$(jq -n \
    --arg model "$MODEL" \
    --arg temp "$TEMPERATURE" \
    --arg prompt "$prompt" \
    ' {
      model: $model,
      temperature: ($temp | tonumber),
      messages: [
        { role: "system", content: "You are a senior engineer performing code review." },
        { role: "user", content: $prompt }
      ]
    }')" |
jq -r '.choices[0].message.content'

# vim:tabstop=2:softtabstop=2:shiftwidth=2:expandtab:filetype=sh

RAG

[!TIP]
RAG - Retrieval-Augmented Generation

tips

chatgpt

see details

import os
from pathlib import Path

# Define script contents
table_script = """#!/usr/bin/env bash

# print_table.sh — CLI 表格打印工具，支持列对齐和彩色输出

_strip_ansi() {
  sed 's/\\x1B\\[[0-9;]*[mK]//g'
}

_str_length() {
  echo -e "$1" | _strip_ansi | awk '{ print length }'
}

repeat_char() {
  printf "%*s" "$2" '' | tr ' ' "$1"
}

print_table() {
  local -n _headers=$1
  local -n _rows=$2
  local -n _aligns=$3

  local cols=${#_headers[@]}
  local -a widths

  # 计算每列最大宽度
  for ((i = 0; i < cols; i++)); do
    widths[i]=$(_str_length "${_headers[i]}")
  done

  for row in "${_rows[@]}"; do
    IFS='|' read -r -a fields <<<"$row"
    for ((i = 0; i < cols; i++)); do
      len=$(_str_length "${fields[i]}")
      ((len > widths[i])) && widths[i]=$len
    done
  done

  # 打印分割线
  _print_line() {
    for ((i = 0; i < cols; i++)); do
      printf "+-%s" "$(repeat_char '-' "${widths[i]}")"
    done
    echo "+"
  }

  # 打印一行
  _print_row() {
    local -a fields=("$@")
    for ((i = 0; i < cols; i++)); do
      local raw="${fields[i]}"
      local clean=$(_str_length "$raw")
      local pad=${widths[i]}
      case "${_aligns[i]}" in
        right)
          printf "| %*s " "$pad" "$raw"
          ;;
        center)
          local left=$(( (pad - clean) / 2 ))
          local right=$(( pad - clean - left ))
          printf "| %*s%s%*s " "$left" "" "$raw" "$right" ""
          ;;
        *)
          printf "| %-*s " "$pad" "$raw"
          ;;
      esac
    done
    echo "|"
  }

  # 打印整张表格
  _print_line
  _print_row "${_headers[@]}"
  _print_line
  for row in "${_rows[@]}"; do
    IFS='|' read -r -a fields <<<"$row"
    _print_row "${fields[@]}"
  done
  _print_line
}
"""

# Define output path
output_path = Path("/mnt/data/cli-ui/print_table.sh")
output_path.parent.mkdir(parents=True, exist_ok=True)

# Save script to file
with open(output_path, "w") as f:
    f.write(table_script)

output_path.name

Previouspodman Nextprompt

Last updated 4 months ago

Was this helpful?