AI Models

The Heroku Managed Inference and Agent add-on supports the following models. The add-on is hosted in two regions: us and eu. However, the add-on can be provisioned and accessed from apps in any Heroku region. Select a model to view information on rate limits, prompt caching, and implementation.

Model Documentation Model ID Region Supported Inputs Supported Outputs API Endpoint Model Source Description
Amazon Rerank 1.0 amazon-rerank-1-0 US, EU text score /v1/rerank Amazon A reliable, high-performing reranking model backed by AWS infrastructure.
Nova Lite nova-lite US, EU text, image, video text /v1/chat/completions Amazon A fast and cost-effective LLM.
Nova 2 Lite nova-2-lite US, EU text, image, video text /v1/chat/completions Amazon A fast and cost-effective LLM that supports conversational chat, tool-calling, and advanced reasoning with extended context.
Nova Pro nova-pro US, EU text, image, video text /v1/chat/completions Amazon A high-performance LLM designed for complex tasks.
Claude 4.5 Haiku claude-4-5-haiku US, EU text, image text /v1/chat/completions Anthropic A state-of-the-art LLM that supports chat, tool-calling, and enhanced reasoning.
Claude Haiku 5.5 claude-haiku-5-5 US, EU text, image text /v1/chat/completions Anthropic A fast and cost-effective LLM designed for high-volume, latency-sensitive tasks like classification, extraction, and routing.
Claude 4.5 Sonnet claude-4-5-sonnet US, EU text, image text /v1/chat/completions Anthropic A state-of-the-art LLM optimized for enterprise apps that supports chat, tool-calling, and enhanced reasoning.
Claude Sonnet 4.6 claude-sonnet-4-6 US, EU text, image text /v1/chat/completions Anthropic A state-of-the-art LLM designed for complex tasks including data processing, sales forecasting, and content generation.
Claude Sonnet 5.5 claude-sonnet-5-5 US, EU text, image text /v1/chat/completions Anthropic A state-of-the-art LLM designed for data processing, nuanced content generation, and real-time apps that demand high endurance and quality.
Claude Opus 4.5 claude-opus-4-5 US, EU text, image text /v1/chat/completions Anthropic A next-generation, frontier LLM that supports chat, tool-calling, autonomous coding, effort control, and enhanced reasoning.
Claude Opus 4.6 claude-opus-4-6 US, EU text, image text /v1/chat/completions Anthropic A next-generation, frontier LLM that supports chat, tool-calling, autonomous coding, effort control, and enhanced reasoning.
Claude Opus 4.7 claude-opus-4-7 US, EU text, image text /v1/chat/completions Anthropic A next-generation, frontier LLM that supports chat, tool-calling, autonomous coding, effort control, and enhanced reasoning.
Claude Opus 4.8 claude-opus-4-8 US, EU text, image text /v1/chat/completions Anthropic A next-generation, frontier LLM that supports chat, tool-calling, autonomous coding, effort control, and enhanced reasoning.
Claude Opus 5.5 claude-opus-5-5 US, EU text, image text /v1/chat/completions Anthropic A next-generation, frontier LLM designed for long-running agents with deeper reasoning for advanced coding, complex professional work, and multi-step agent tasks.
Claude Fable 5.1 claude-fable-5-1 US, EU text, image text /v1/chat/completions Anthropic A state-of-the-art LLM designed for the most demanding reasoning and long-horizon agentic work that require advanced coding, extended multi-step tasks, and complex analysis.
Cohere Embed Multilingual cohere-embed-multilingual US, EU text embedding /v1/embeddings Cohere A state-of-the-art embedding model that supports multiple languages and can be helpful for developing RAG search.
Cohere Embed V4 cohere-embed-v4 US, EU text embedding /v1/embeddings Cohere A state-of-the-art embedding model that supports over 100 languages and can be helpful for developing RAG search.
Cohere Rerank 3.5 cohere-rerank-3-5 US, EU text score /v1/rerank Cohere A reranking model that offers enhanced reasoning, broad data compatibility, and multilingual support.
OpenAI gpt-oss-120b gpt-oss-120b US, EU text text /v1/chat/completions OpenAI An open-weight LLM that supports chat and tool-calling.
Stable Image Core stable-image-core US text image /v1/images/generations Stability AI A fast and cost-effective diffusion (image generation) model.
Stable Image Ultra stable-image-ultra US text image /v1/images/generations Stability AI A state-of-the-art diffusion (image generation) model.

Deprecated Models

Heroku is deprecating the following models and will each end-of-life on the dates listed. During the deprecation period, requests to these models return a warning header. Before the EOL date, we’ll convert model-specific plans for deprecated models to the standard plan. After the EOL date, requests to these models return HTTP 410.

Model Model ID Deprecation Date EOL Date Replacement
Claude 3.5 Sonnet Latest claude-3-5-sonnet-latest January 22, 2026 February 22, 2026 claude-4-6-sonnet
Claude 3.7 Sonnet claude-3-7-sonnet March 21, 2026 April 21, 2026 claude-4-6-sonnet
Claude 3.5 Haiku claude-3-5-haiku May 12, 2026 June 12, 2026 claude-4-5-haiku
DeepSeek V3.2 deepseek-v3-2 June 15, 2026 July 15, 2026 gpt-oss-120b
MiniMax M2 minimax-m2 June 15, 2026 July 15, 2026 gpt-oss-120b
MiniMax M2.1 minimax-m2-1 June 15, 2026 July 15, 2026 gpt-oss-120b
Kimi K2 Thinking kimi-k2-thinking June 15, 2026 July 15, 2026 gpt-oss-120b
Kimi K2.5 kimi-k2-5 June 15, 2026 July 15, 2026 gpt-oss-120b
Qwen3 235B qwen3-235b June 15, 2026 July 15, 2026 gpt-oss-120b
Qwen3 Coder 480B qwen3-coder-480b June 15, 2026 July 15, 2026 gpt-oss-120b
GLM 4.7 glm-4-7 June 15, 2026 July 15, 2026 gpt-oss-120b
GLM 4.7 Flash glm-4-7-flash June 15, 2026 July 15, 2026 gpt-oss-120b
Claude 3 Haiku claude-3-haiku August 3, 2026 September 3, 2026 claude-4-5-haiku
Claude 4 Sonnet claude-4-sonnet September 7, 2026 October 7, 2026 claude-4-6-sonnet