ibm-granite/granite-4.2-8b

ibm-granite/granite-4.2-8b is a budget-tier language model from RouterPlus. ibm-granite/granite-4.2-8b is an open-weights model, meaning the weights are publicly available for self-hosting, fine-tuning, and deployment on your own infrastructure — not just through RouterPlus's API. The Granite family is RouterPlus's model series for this capability tier.

ibm-granite/granite-4.2-8b is priced at $0.06/1M input tokens and $0.25/1M output tokens. Output tokens are 4.2× more expensive than input tokens — this ratio reflects the sequential nature of autoregressive generation, which requires a separate forward pass per output token versus parallel processing of all input tokens simultaneously. At a typical usage pattern of 1,000 input tokens and 500 output tokens per call, ibm-granite/granite-4.2-8b costs approximately $0.18 per 1,000 API calls. Prompt caching reduces repeated input tokens to $0.015/1M — a 75% discount that compounds significantly for applications with long shared system prompts, multi-turn conversations, or document Q&A where the same context is reused across many queries.

Its 131K-token context window handles most long-document tasks — roughly 75,000 words or a full novel — without requiring chunking or retrieval augmentation. Maximum output length is 131K tokens — sufficient for long-form content generation, detailed code files, and extended reports.

ibm-granite/granite-4.2-8b is best suited to general language tasks. Supported capabilities include streaming token output for real-time UIs. When evaluating this model for a specific workload, the most important factors are: context window (does your input fit?), output price (how long are your generations?), and latency (is this interactive or batch?). The pricing table above shows how ibm-granite/granite-4.2-8b compares to similar models across these dimensions.

ibm-granite/granite-4.2-8b sits in the budget tier — the most cost-efficient models, best for high-volume simple tasks, classification, extraction, and applications where cost per call is the primary constraint. As an open-weights model, ibm-granite/granite-4.2-8b can be self-hosted on GPU compute (see the GPU rental comparison on this site) to eliminate per-token API costs entirely — the break-even point versus API pricing depends on your volume and the GPU hourly rate. When comparing ibm-granite/granite-4.2-8b to alternatives, use the similar models section on this page to find models at comparable price points with different capability tradeoffs. The intelligence score (N/A/100) reflects aggregate benchmark performance — a useful proxy for general capability, though real-world performance varies significantly by task type. Always test with your specific prompts and use case before committing to a model for production.