watsonx.ai
5 modelsProvider docs
As of Nov 6, 2026, watsonx.ai lists 5 supported models in the Sovyron catalog, including Granite-4.0-H-Small, GPT OSS 120B, Llama 4 Maverick 17B 128E Instruct FP8, Mistral Small 3.1 24B and Llama-3.3-70B-Instruct across the granite, gpt-oss, llama and mistral-small families. The cheapest rate in the table is Granite-4.0-H-Small at $0.265 per 1M output tokens, the priciest Llama 4 Maverick 17B 128E Instruct FP8 at $1.484. On a 3:1 input:output blend the metered rates span $0.114–$0.7526 per 1M tokens, and every price is watsonx.ai's own, normalized to USD. Offer mix: metered per-token rates only, with no free-tier or subscription offers in the catalog. Open a model to compare it with the cheapest offer across all providers.
Supported models and pricing
5 of 5 models
| Model | 1M input | 1M output | 1M cache read | Context | Released |
|---|---|---|---|---|---|
| Granite-4.0-H-Small | $0.0636 | $0.265 | — | 131k | Oct 2, 2025 |
| GPT OSS 120B | $0.159 | $0.636 | — | 131k | Aug 5, 2025 |
| Llama 4 Maverick 17B 128E Instruct FP8 | $0.371 | $1.484 | — | 131k | Apr 5, 2025 |
| Mistral Small 3.1 24B | $0.106 | $0.318 | — | 131k | Mar 17, 2025 |
| Llama-3.3-70B-Instruct | $0.7526 | $0.7526 | — | 131k | Dec 6, 2024 |