Supported Models

Overview

SaveGate provides access to 400+ AI models from leading providers. All models are accessible through a unified API using the same authentication and request format. Pricing shown is per 1M tokens (input/output).

OpenAI Models

Chat Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`gpt-5.4`	1.1M	128K	$2.50	$15
`gpt-5.4-2026-03-05`	1.1M	128K	$2.50	$15
`gpt-5.4-mini`	272K	128K	$0.75	$4.50
`gpt-5.4-nano`	272K	128K	$0.20	$1.25
`gpt-5.3-chat-latest`	128K	16.4K	$1.75	$14
`gpt-5.2`	272K	128K	$1.75	$14
`gpt-5.2-chat-latest`	128K	16.4K	$1.75	$14
`gpt-5.2-2025-12-11`	272K	128K	$1.75	$14
`gpt-5.1`	272K	128K	$1.25	$10
`gpt-5.1-chat-latest`	128K	16.4K	$1.25	$10
`gpt-5.1-2025-11-13`	272K	128K	$1.25	$10
`gpt-5`	272K	128K	$1.25	$10
`gpt-5-chat`	128K	16.4K	$1.25	$10
`gpt-5-chat-latest`	128K	16.4K	$1.25	$10
`gpt-5-search-api`	272K	128K	$1.25	$10
`gpt-5-2025-08-07`	272K	128K	$1.25	$10
`gpt-5-search-api-2025-10-14`	272K	128K	$1.25	$10
`gpt-5-mini`	272K	128K	$0.25	$2
`gpt-5-mini-2025-08-07`	272K	128K	$0.25	$2
`gpt-5-nano`	272K	128K	$0.05	$0.40
`gpt-5-nano-2025-08-07`	272K	128K	$0.05	$0.40
`o4-mini`	200K	100K	$1.10	$4.40
`o4-mini-2025-04-16`	200K	100K	$1.10	$4.40
`chatgpt-4o-latest`	128K	4.1K	$5	$15
`gpt-4o`	128K	16.4K	$2.50	$10
`gpt-4o-search-preview`	128K	16.4K	$2.50	$10
`gpt-4o-2024-05-13`	128K	4.1K	$5	$15
`gpt-4o-2024-08-06`	128K	16.4K	$2.50	$10
`gpt-4o-2024-11-20`	128K	16.4K	$2.50	$10
`gpt-4o-search-preview-2025-03-11`	128K	16.4K	$2.50	$10
`gpt-4o-mini`	128K	16.4K	$0.15	$0.60
`gpt-4o-mini-search-preview`	128K	16.4K	$0.15	$0.60
`gpt-4o-mini-2024-07-18`	128K	16.4K	$0.15	$0.60
`gpt-4o-mini-search-preview-2025-03-11`	128K	16.4K	$0.15	$0.60
`gpt-4-turbo`	128K	4.1K	$10	$30
`gpt-4.1`	1.0M	32.8K	$2	$8
`gpt-4-turbo-2024-04-09`	128K	4.1K	$10	$30
`gpt-4.1-2025-04-14`	1.0M	32.8K	$2	$8
`gpt-4.1-mini`	1.0M	32.8K	$0.40	$1.60
`gpt-4.1-mini-2025-04-14`	1.0M	32.8K	$0.40	$1.60
`gpt-4.1-nano`	1.0M	32.8K	$0.10	$0.40
`gpt-4.1-nano-2025-04-14`	1.0M	32.8K	$0.10	$0.40
`gpt-4`	8.2K	4.1K	$30	$60
`gpt-4-0613`	8.2K	4.1K	$30	$60
`o3`	200K	100K	$2	$8
`o3-2025-04-16`	200K	100K	$2	$8
`o3-mini`	200K	100K	$1.10	$4.40
`o3-mini-2025-01-31`	200K	100K	$1.10	$4.40
`gpt-3.5-turbo`	16.4K	4.1K	$0.50	$1.50
`o1`	200K	100K	$15	$60
`o1-2024-12-17`	200K	100K	$15	$60
`gpt-realtime`	32K	4.1K	$4	$16
`gpt-realtime-1.5`	32K	4.1K	$4	$16
`gpt-realtime-2025-08-28`	32K	4.1K	$4	$16
`gpt-realtime-mini`	128K	4.1K	$0.60	$2.40
`gpt-realtime-mini-2025-10-06`	128K	4.1K	$0.60	$2.40
`gpt-realtime-mini-2025-12-15`	128K	4.1K	$0.60	$2.40

Responses Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`gpt-5.4-pro`	1.1M	128K	$30	$180
`gpt-5.4-pro-2026-03-05`	1.1M	128K	$30	$180
`gpt-5.3-codex`	272K	128K	$1.75	$14
`gpt-5.2-pro`	272K	128K	$21	$168
`gpt-5.2-pro-2025-12-11`	272K	128K	$21	$168
`gpt-5.2-codex`	272K	128K	$1.75	$14
`gpt-5.1-codex`	272K	128K	$1.25	$10
`gpt-5.1-codex-max`	272K	128K	$1.25	$10
`gpt-5.1-codex-mini`	272K	128K	$0.25	$2
`gpt-5-pro`	128K	272K	$15	$120
`gpt-5-pro-2025-10-06`	128K	272K	$15	$120
`gpt-5-codex`	272K	128K	$1.25	$10
`o4-mini-deep-research`	200K	100K	$2	$8
`o4-mini-deep-research-2025-06-26`	200K	100K	$2	$8
`o3-pro`	200K	100K	$20	$80
`o3-pro-2025-06-10`	200K	100K	$20	$80
`o3-deep-research`	200K	100K	$10	$40
`o3-deep-research-2025-06-26`	200K	100K	$10	$40
`o1-pro`	200K	100K	$150	$600
`o1-pro-2025-03-19`	200K	100K	$150	$600
`codex-mini-latest`	200K	100K	$1.50	$6

Embedding Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`text-embedding-3-large`	8.2K	-	$0.13	-
`text-embedding-3-small`	8.2K	-	$0.02	-
`text-embedding-ada-002`	8.2K	-	$0.10	-
`text-embedding-ada-002-v2`	8.2K	-	$0.10	-

Image Generation Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`chatgpt-image-latest`	-	-	$5	-
`standard/1024-x-1024/dall-e-3`	-	-	-	-
`standard/1024-x-1792/dall-e-3`	-	-	-	-
`standard/1792-x-1024/dall-e-3`	-	-	-	-
`gpt-image-1`	-	-	$5	-
`gpt-image-1.5`	-	-	$5	$10
`gpt-image-1.5-2025-12-16`	-	-	$5	$10
`gpt-image-1-mini`	-	-	$2	-

Image generation models also support resolution and quality variants (e.g., hd/1024-x-1024/dall-e-3). Use the /v1/models API endpoint to see all available variants.

Audio Speech Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`gpt-4o-mini-tts`	-	-	$2.50	$10
`gpt-4o-mini-tts-2025-03-20`	-	-	$2.50	$10
`gpt-4o-mini-tts-2025-12-15`	-	-	$2.50	$10
`tts-1`	-	-	-	-
`tts-1-1106`	-	-	-	-
`tts-1-hd`	-	-	-	-
`tts-1-hd-1106`	-	-	-	-

Audio Transcription Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`gpt-4o-transcribe`	16K	2K	$2.50	$10
`gpt-4o-transcribe-diarize`	16K	2K	$2.50	$10
`gpt-4o-mini-transcribe`	16K	2K	$1.25	$5
`gpt-4o-mini-transcribe-2025-03-20`	16K	2K	$1.25	$5
`gpt-4o-mini-transcribe-2025-12-15`	16K	2K	$1.25	$5
`whisper-1`	-	-	-	-

Video Generation Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`sora-2-pro`	-	-	-	-
`sora-2-pro-high-res`	-	-	-	-
`sora-2`	-	-	-	-

Completion Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`babbage-002`	16.4K	4.1K	$0.40	$0.40

Moderation Models

Model ID	Max Input	Input/1M	Output/1M
`omni-moderation-latest`	32.8K	-	-
`text-moderation-007`	32.8K	-	-
`text-moderation-latest`	32.8K	-	-
`text-moderation-stable`	32.8K	-	-
`omni-moderation-2024-09-26`	32.8K	-	-

Anthropic Claude Models

Chat Models

Model ID	Max Input	Max Output	Input/1M	Output/1M
`claude-opus-4-7`	1M	128K	$5	$25
`claude-opus-4-6`	1M	128K	$5	$25
`claude-sonnet-4-6`	1M	64K	$3	$15
`claude-opus-4-5`	200K	64K	$5	$25
`claude-opus-4-5-20251101`	200K	64K	$5	$25
`claude-sonnet-4-5`	200K	64K	$3	$15
`claude-sonnet-4-5-20250929`	200K	64K	$3	$15
`claude-haiku-4-5`	200K	64K	$1	$5
`claude-haiku-4-5-20251001`	200K	64K	$1	$5
`claude-opus-4-20250514`	200K	32K	$15	$75
`claude-sonnet-4-20250514`	1M	64K	$3	$15
`claude-opus-4-1`	200K	32K	$15	$75
`claude-opus-4-1-20250805`	200K	32K	$15	$75
`claude-4-opus-20250514`	200K	32K	$15	$75
`claude-4-sonnet-20250514`	1M	64K	$3	$15
`claude-3-haiku-20240307`	200K	4.1K	$0.25	$1.25

Google Gemini Models

Chat Models

Chat models are accessed via the standard OpenAI-compatible endpoint. Use the model IDs below directly — no prefix needed.

Model ID	Max Input	Max Output	Input/1M	Output/1M
`gemini-3.7-flash`	1.0M	64K	$0.75	$3.75
`gemini-2.5-pro`	1.0M	64K	$1.25	$10
`gemini-2.5-flash`	1.0M	64K	$0.30	$2.50

Controlling thinking cost

Gemini models think before answering by default, and those reasoning tokens are billed as output. On short tasks — classification, extraction, formatting — that can be most of your bill. Use reasoning_effort to turn it down. The accepted value differs per model, which is easy to get wrong:

Model	Cheapest value	Notes
`gemini-2.5-flash`	`"none"`	Reduces reasoning to 0 tokens
`gemini-2.5-pro`	`"minimal"`	`"none"` returns 400 — this model cannot disable thinking
`gemini-3.7-flash`	`"low"`	`"none"` and `"minimal"` both return 400

curl https://api.savegate.ai/v1/chat/completions \
  -H "Authorization: Bearer $SAVEGATE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.7-flash",
    "reasoning_effort": "low",
    "messages": [{"role": "user", "content": "Extract the city: I flew to Da Nang on Friday."}]
  }'

Leave reasoning_effort unset when answer quality matters more than cost.

Image Generation Models

Google calls these models “nano banana” and “nano banana pro”. The IDs below are the official API names — use those.

Model ID	Input/1M	Output/1M	Per generated image
`gemini-2.5-flash-image`	$0.30	$2.50	$0.039
`gemini-3-pro-image`	$2.00	$12.00	$0.134

Generated images are billed at a flat rate per image —

0.039 for `gemini-2.5-flash-image`,

0.134 for gemini-3-pro-image — regardless of resolution. Text in the same request (your prompt, and any text the model returns) is billed separately at the input/output rates above. Both figures appear in your usage report.

Both models work through the standard OpenAI-compatible endpoint. Two ways to call them:

1. `/v1/images/generations` — plain generation

curl https://api.savegate.ai/v1/images/generations \
  -H "Authorization: Bearer $SAVEGATE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-2.5-flash-image",
    "prompt": "A cozy Vietnamese coffee shop interior, warm morning light, phin filter on a wooden table, photorealistic"
  }' \
| jq -r '.data[0].b64_json' | base64 --decode > output.png

The response contains data[0].b64_json, not data[0].url. You must decode the base64 yourself — there is no hosted image link.

gemini-2.5-flash-image · 1024×1024 · ~$0.058

Set the frame shape with imageConfig:

{
  "model": "gemini-3-pro-image",
  "prompt": "...",
  "imageConfig": { "aspectRatio": "16:9" }
}

`aspectRatio`	Output size
(omitted)	1024×1024
`16:9`	1408×768

gemini-3-pro-image · 16:9 → 1408×768 · ~$0.202

2. `/v1/chat/completions` — editing, or text + image together

Send an image in and describe the change. This is the only way to edit an image.

# Build the request in a FILE. A base64-encoded image is 1-2 MB, which overflows
# the shell argument limit if you inline it — you would get
# "argument list too long" rather than an API error.
base64 -i input.png | tr -d '\n' > b64.txt

jq -n --rawfile b64 b64.txt '{
  model: "gemini-3-pro-image",
  modalities: ["image", "text"],
  messages: [{
    role: "user",
    content: [
      {type: "text", text: "Make the mug bright red. Keep everything else the same."},
      {type: "image_url", image_url: {url: ("data:image/png;base64," + $b64)}}
    ]
  }]
}' > req.json

curl https://api.savegate.ai/v1/chat/completions \
  -H "Authorization: Bearer $SAVEGATE_API_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @req.json \
| jq -r '.choices[0].message.images[0].image_url.url' \
| sed 's|^data:image/[^;]*;base64,||' | base64 --decode > edited.png

On this path the image is at choices[0].message.images[0].image_url.url as a data URI — note images, plural. Reading message.content returns null.

Practical notes

A response is 1–2 MB of base64. Set a generous client timeout (60s+) and never print it straight to a terminal.
When sending an image in, build the JSON in a file (--data-binary @req.json). Inlining base64 into a shell argument fails with argument list too long.
n (multiple images per request) is not supported — issue separate requests.
Choose gemini-2.5-flash-image unless you need the quality: gemini-3-pro-image costs about 3.5× more per image.

DeepSeek Models

Chat Models

Model ID	Tier	Uncached Input/1M	Cached Input/1M	Output/1M
`deepseek-v4-pro`	Standard	$1.74	$0.145	$3.48

Standard tier — for general use, balancing efficient performance and price.Cached input applies to prompt prefixes that hit the cache on repeat requests, at roughly 1/12 the uncached input price.

ElevenLabs Models

Text-to-Speech Models

Model ID	Mode	Pricing
`elevenlabs/eleven_v3`	Audio Speech	$0.18 / 1K characters

Speech-to-Text Models

Model ID	Mode	Pricing
`elevenlabs/scribe_v1_experimental`	Audio Transcription	$0.22 / hour

Using Models

Simply specify the model ID in your API request:

response = client.chat.completions.create(
    model="gpt-5.4",  # Specify model here
    messages=[{"role": "user", "content": "Hello!"}]
)

const response = await client.chat.completions.create({
  model: 'claude-sonnet-4-6',  // Specify model here
  messages: [{ role: 'user', content: 'Hello!' }]
});

Model Updates

SaveGate automatically updates model versions to the latest stable releases. For version pinning, use specific model IDs (with date suffix) when available.

Use the /v1/models API endpoint to get the full list of available models with their capabilities and pricing information in real-time.

Getting Started

Core Concepts

SDK Integration

Guides

Overview

OpenAI Models

Chat Models

Responses Models

Embedding Models

Image Generation Models

Audio Speech Models

Audio Transcription Models

Video Generation Models

Completion Models

Moderation Models

Anthropic Claude Models

Chat Models

Google Gemini Models

Chat Models

Controlling thinking cost

Image Generation Models

1. `/v1/images/generations` — plain generation

2. `/v1/chat/completions` — editing, or text + image together

Practical notes

DeepSeek Models

Chat Models

ElevenLabs Models

Text-to-Speech Models

Speech-to-Text Models

Using Models

Model Updates

​Overview

​OpenAI Models

​Chat Models

​Responses Models

​Embedding Models

​Image Generation Models

​Audio Speech Models

​Audio Transcription Models

​Video Generation Models

​Completion Models

​Moderation Models

​Anthropic Claude Models

​Chat Models

​Google Gemini Models

​Chat Models

​Controlling thinking cost

​Image Generation Models

​1. /v1/images/generations — plain generation

​2. /v1/chat/completions — editing, or text + image together

​Practical notes

​DeepSeek Models

​Chat Models

​ElevenLabs Models

​Text-to-Speech Models

​Speech-to-Text Models

​Using Models

​Model Updates

Overview

OpenAI Models

Chat Models

Responses Models

Embedding Models

Image Generation Models

Audio Speech Models

Audio Transcription Models

Video Generation Models

Completion Models

Moderation Models

Anthropic Claude Models

Chat Models

Google Gemini Models

Chat Models

Controlling thinking cost

Image Generation Models

1. `/v1/images/generations` — plain generation

2. `/v1/chat/completions` — editing, or text + image together

Practical notes

DeepSeek Models

Chat Models

ElevenLabs Models

Text-to-Speech Models

Speech-to-Text Models

Using Models

Model Updates