New models and precision tool calling endpoints

New models and precision tool calling endpoints

Hi vv,

We are introducing :exacto - Precision Tool-Calling Endpoints.

These endpoints are focused on higher tool-calling accuracy by routing requests to providers that demonstrate measurably better tool calling performance.

Exacto endpoints are available for:
- Kimi K2
- DeepSeek v3.1 Terminus
- GLM 4.6 
- GPT‑OSS 120B 
- Qwen3 Coder 

While model weights are identical across providers, real-world inference quality can differs. OpenRouter sees billions of requests per month, giving us a unique view into how models behave across inference stacks and allowing us to curate the most accurate providers for agentic and tool-heavy workloads.

For example - If you are using MCPs, you should try the :exacto variant for better tool calling accuracy. 

Read more about this in the blog post or the documentation.

Next up, we have launched two new image models and a stealth model -

  • GPT 5 - Image and GPT 5 - Image mini
  • Andromeda Alpha is the new stealth model. It is a small reasoning model that is optimized for image understanding. All prompts and responses will be logged by the provider to improve their services.

Best,
OpenRouter Team
For questions or comments, reach out to us on X or on Discord.

Comments

Popular posts from this blog

Make quick edits right in your PDF 📝

Strike the right chord with a PDF 🎵

Pitch perfect PDFs 💯