Models

Every major LLM through GammaInfra's smart router. Pricing is pass-through — provider rates with a 3% top-up fee during the launch window (5% standard), no per-token markup. Click a model for the full spec sheet.

Routing options:

OpenAI

SlugInput / 1MOutput / 1MContextToolsVisionJSON
openai/gpt-5$1.25$10.00256K
openai/gpt-5-mini$0.25$2.00256K
openai/gpt-5-nano$0.05$0.40256K
openai/gpt-5.5$5.00$30.00272K

Anthropic

SlugInput / 1MOutput / 1MContextToolsVisionJSON
anthropic/claude-opus-4-7$5.00$25.00200K
anthropic/claude-opus-4-6$5.00$25.00200K
anthropic/claude-sonnet-4-6$3.00$15.00200K
anthropic/claude-haiku-4-5$0.80$4.00200K

Google

SlugInput / 1MOutput / 1MContextToolsVisionJSON
google/gemini-3.1-pro-preview$2.50$15.002M
google/gemini-3-flash-preview$0.30$2.501M
google/gemini-3.1-flash-lite-preview$0.07$0.301M

DeepSeek

SlugInput / 1MOutput / 1MContextToolsVisionJSON
deepseek/deepseek-v4-pro$0.50$2.00128K
deepseek/deepseek-v4-flash$0.14$0.55128K

Mistral

SlugInput / 1MOutput / 1MContextToolsVisionJSON
mistral/mistral-large-2512$2.00$6.00128K
mistral/mistral-small-2603$0.20$0.60128K
mistral/devstral-2512$0.40$1.60128K

Groq (Llama)

SlugInput / 1MOutput / 1MContextToolsVisionJSON
groq/llama-3.3-70b-versatile$0.59$0.79128K
groq/llama-3.1-8b-instant$0.05$0.08128K

xAI (Grok)

SlugInput / 1MOutput / 1MContextToolsVisionJSON
grok/grok-4-1-fast-non-reasoning$0.30$1.20256K

Amazon Bedrock

SlugInput / 1MOutput / 1MContextToolsVisionJSON
bedrock/us.anthropic.claude-opus-4-7$5.00$25.00200K
bedrock/us.amazon.nova-pro-v1:0$0.80$3.20300K

Frequently asked questions

Which models can I call through GammaInfra?
Models from every major provider — OpenAI, Anthropic, Google, Mistral, Groq, DeepSeek, xAI, and Amazon Bedrock. Browse the catalog for per-model pricing and capabilities. Pin any of them directly with a provider/model name, or use gammainfra/auto and let the router pick per prompt.
How is model pricing shown?
Each model page lists the provider pass-through input and output rate per 1M tokens. GammaInfra adds 0% markup on tokens; its fee is taken at top-up (3% launch / 5% standard) or 1–2% per request on BYOK. Every response also returns the exact spend in X-GammaInfra-Cost-USD.
Do I have to pick a model?
No. gammainfra/auto classifies each prompt by task and routes it to the best-fit model automatically, with cross-provider fallback. Pin a specific model only when the choice is part of your application's design rather than a routing variable.
How current is the model catalog?
The catalog tracks the providers' shipping models; new models are added as provider adapters are wired in. Pricing and capability fields on each model page are maintained against the providers' published rates — the X-GammaInfra-Cost-USD response header is always authoritative for actual spend.