Supported Models

SandBase provides access to 1400+ models from 50+ providers through a single API. All models are available via the OpenAI-compatible endpoint (/v1/chat/completions) and, where applicable, the Anthropic-compatible endpoint (/v1/messages).

TIP

Browse the full model catalog with live pricing at sandbase.ai/models.

Model Comparison Table

Model	Provider	Input (per 1M tokens)	Output (per 1M tokens)	Context Window	Key Capabilities
openai/gpt-4o	OpenAI	$2.50	$10.00	128K	Chat, Streaming, Tools, Vision, JSON Mode, Cache
openai/gpt-4o-mini	OpenAI	$0.15	$0.60	128K	Chat, Streaming, Tools, Vision, JSON Mode, Cache
openai/o3	OpenAI	$10.00	$40.00	200K	Chat, Streaming, Tools, Vision, Thinking, JSON Mode
openai/o3-mini	OpenAI	$1.10	$4.40	200K	Chat, Streaming, Tools, Thinking, JSON Mode
anthropic/claude-sonnet-4	Anthropic	$3.00	$15.00	200K	Chat, Streaming, Tools, Vision, Thinking, Cache
anthropic/claude-3.5-haiku	Anthropic	$0.80	$4.00	200K	Chat, Streaming, Tools, Vision, Cache
deepseek/deepseek-v3	DeepSeek	$0.27	$1.10	128K	Chat, Streaming, Tools, JSON Mode
deepseek/deepseek-chat	DeepSeek	$0.27	$1.10	128K	Chat, Streaming, Tools, JSON Mode
deepseek/deepseek-reasoner	DeepSeek	$0.55	$2.19	128K	Chat, Streaming, Thinking
google/gemini-2.5-pro	Google	$1.25	$10.00	1M	Chat, Streaming, Tools, Vision, Thinking, JSON Mode, Cache
google/gemini-2.5-flash	Google	$0.15	$0.60	1M	Chat, Streaming, Tools, Vision, Thinking, JSON Mode, Cache
meta/llama-4-maverick	Meta	$0.20	$0.60	128K	Chat, Streaming, Tools, Vision
alibaba/qwen3-32b	Alibaba	$0.10	$0.30	32K	Chat, Streaming, Tools
bytedance/seed-1.6	ByteDance	$0.50	$2.00	128K	Chat, Streaming, Tools, Vision

TIP

Prices are in USD and reflect SandBase's pass-through pricing. Actual costs may vary slightly based on provider pricing changes.

Models by Provider

OpenAI

Model	Best For	Context	Reasoning
openai/gpt-4o	General-purpose, multimodal tasks	128K	No
openai/gpt-4o-mini	Simple tasks, high-volume workloads	128K	No
openai/o3	Complex reasoning, math, coding	200K	Yes
openai/o3-mini	Budget reasoning tasks	200K	Yes

Notes:

• o3 and o3-mini support reasoning_effort parameter (low, medium, high)
• Reasoning tokens are billed at the output token rate
• All OpenAI models support automatic prompt caching (50% discount on cache hits)

Anthropic

Model	Best For	Context	Reasoning
anthropic/claude-sonnet-4	Coding, analysis, complex instructions	200K	Yes (extended thinking)
anthropic/claude-3.5-haiku	Fast responses, simple tasks	200K	No

Notes:

• Claude Sonnet 4 supports extended thinking with configurable budget_tokens
• Anthropic models support explicit prompt caching with cache_control (90% discount on reads)
• Available via both /v1/chat/completions and /v1/messages endpoints

Google

Model	Best For	Context	Reasoning
google/gemini-2.5-pro	Long-context tasks, multimodal	1M	Yes
google/gemini-2.5-flash	Fast, cost-effective general use	1M	Yes

Notes:

• 1 million token context window — largest available
• Both models support thinking/reasoning
• Gemini 2.5 Flash offers the best price-to-performance ratio for many tasks

DeepSeek

Model	Best For	Context	Reasoning
deepseek/deepseek-v3	General-purpose, coding	128K	No
deepseek/deepseek-chat	Conversational AI	128K	No
deepseek/deepseek-reasoner	Step-by-step reasoning	128K	Yes

Notes:

• Extremely cost-effective — among the cheapest models available
• DeepSeek Reasoner uses chain-of-thought reasoning (similar to o3)
• Automatic prompt caching with 90% discount on cache hits

Meta (Open Source)

Model	Best For	Context	Reasoning
meta/llama-4-maverick	General-purpose, open-source alternative	128K	No

Notes:

• Open-source model hosted by multiple providers
• Good balance of quality and cost
• Supports vision (image understanding)

Alibaba

Model	Best For	Context	Reasoning
alibaba/qwen3-32b	Budget tasks, multilingual	32K	No

Notes:

• Lowest cost model available on SandBase
• Strong multilingual capabilities (especially Chinese)
• 32K context window (smaller than other models)

ByteDance

Model	Best For	Context	Reasoning
bytedance/seed-1.6	Multimodal, creative tasks	128K	No

Notes:

• Strong multimodal capabilities
• Good for creative writing and image understanding

Choosing a Model

By Use Case

Use Case	Recommended	Why
Chat interface	openai/gpt-4o-mini or anthropic/claude-3.5-haiku	Fast, cheap, good quality
Code generation	anthropic/claude-sonnet-4	Best coding performance
Complex reasoning	openai/o3 or google/gemini-2.5-pro	Deep thinking capabilities
Budget reasoning	openai/o3-mini or deepseek/deepseek-reasoner	Reasoning at lower cost
Long documents	google/gemini-2.5-pro or google/gemini-2.5-flash	1M context window
High volume / classification	openai/gpt-4o-mini or alibaba/qwen3-32b	Lowest cost per token
Multimodal (images)	openai/gpt-4o or google/gemini-2.5-pro	Best vision capabilities

By Budget

Budget Level	Models	Typical Cost per 1K requests
Ultra-low	alibaba/qwen3-32b, openai/gpt-4o-mini	$0.01–$0.05
Low	deepseek/deepseek-v3, google/gemini-2.5-flash	$0.05–$0.20
Medium	anthropic/claude-3.5-haiku, openai/gpt-4o	$0.20–$2.00
High	anthropic/claude-sonnet-4, openai/o3	$2.00–$20.00

Model Availability

All models are available 24/7 through SandBase. If a specific provider experiences downtime, SandBase's routing system automatically falls back to alternative providers when available.

Check real-time model status on the SandBase Status Page.