ModelClock
Capability change
Capability change announced Oct 5, 2026
Tempr Gateway documents a capability change: “Tool calls on gpt-6-luna, gpt-6-sol and gpt-6.1-sol — OpenAI's chat completions rejects function tools on these models unless reasoning is off, and rejects…”
- Hosting route
- Tempr Gateway platform
- Affected scope
- Not stated in the notice
- Announced
- Oct 5, 2026
- First seen by ModelClock
- Oct 6, 2026
What the provider published
temprhq.io ↗Entry date: 2026-10-05
Fixed
- Tool calls on
gpt-6-luna,gpt-6-solandgpt-6.1-sol— OpenAI's chat completions rejects function tools on these models unless reasoning is off, and rejects reasoning effortmaxoutright. Tempr now sends those requests through OpenAI's Responses API, as it already did forgpt-6-astraand thegpt-5.6models, so tools andmaxwork on/v1/chat/completions.
Read from the provider's text; the quote is the provider's exact lines.
