Repository Public Research preview
1// Request builders shared by the live probes and by the linter consistency test.2export const user = [{ role: "user", content: "Reply with the single word: ok" }];3export const tool = { name: "get_time", description: "Return the current time.", input_schema: { type: "object", properties: { tz: { type: "string" } }, required: ["tz"], additionalProperties: false }, strict: true };4export const schema = { type: "json_schema", schema: { type: "object", properties: { answer: { type: "string" } }, required: ["answer"], additionalProperties: false } };56export const PROBES = {7 basic: (m) => ({ model: m, max_tokens: 64, messages: user }),8 thinking_adaptive: (m) => ({ model: m, max_tokens: 512, thinking: { type: "adaptive" }, messages: user }),9 thinking_adaptive_summarized: (m) => ({ model: m, max_tokens: 512, thinking: { type: "adaptive", display: "summarized" }, messages: user }),10 thinking_disabled: (m) => ({ model: m, max_tokens: 64, thinking: { type: "disabled" }, messages: user }),11 thinking_budget: (m) => ({ model: m, max_tokens: 2048, thinking: { type: "enabled", budget_tokens: 1024 }, messages: user }),12 thinking_between_tools: (m) => ({ model: m, max_tokens: 64, thinking: { type: "between_tools" }, messages: user }),13 effort_low: (m) => ({ model: m, max_tokens: 256, output_config: { effort: "low" }, messages: user }),14 effort_medium: (m) => ({ model: m, max_tokens: 256, output_config: { effort: "medium" }, messages: user }),15 effort_high: (m) => ({ model: m, max_tokens: 256, output_config: { effort: "high" }, messages: user }),16 effort_xhigh: (m) => ({ model: m, max_tokens: 256, output_config: { effort: "xhigh" }, messages: user }),17 effort_max: (m) => ({ model: m, max_tokens: 256, output_config: { effort: "max" }, messages: user }),18 disabled_at_xhigh: (m) => ({ model: m, max_tokens: 64, thinking: { type: "disabled" }, output_config: { effort: "xhigh" }, messages: user }),19 temperature: (m) => ({ model: m, max_tokens: 64, temperature: 0.5, messages: user }),20 top_p: (m) => ({ model: m, max_tokens: 64, top_p: 0.9, messages: user }),21 top_k: (m) => ({ model: m, max_tokens: 64, top_k: 40, messages: user }),22 tool_auto: (m) => ({ model: m, max_tokens: 256, tools: [tool], tool_choice: { type: "auto" }, messages: [{ role: "user", content: "What time is it in UTC? Use the tool." }] }),23 tool_any: (m) => ({ model: m, max_tokens: 256, tools: [tool], tool_choice: { type: "any" }, messages: [{ role: "user", content: "What time is it in UTC?" }] }),24 tool_forced: (m) => ({ model: m, max_tokens: 256, tools: [tool], tool_choice: { type: "tool", name: "get_time" }, messages: [{ role: "user", content: "What time is it in UTC?" }] }),25 tool_none: (m) => ({ model: m, max_tokens: 64, tools: [tool], tool_choice: { type: "none" }, messages: user }),26 prefill: (m) => ({ model: m, max_tokens: 64, messages: [...user, { role: "assistant", content: "{" }] }),27 structured_output: (m) => ({ model: m, max_tokens: 256, output_config: { format: schema }, messages: [{ role: "user", content: "Answer in JSON: what is 2+2?" }] }),28 mid_system: (m) => ({ model: m, max_tokens: 64, messages: [...user, { role: "system", content: "Answer in uppercase." }] }),29 cache_control_top: (m) => ({ model: m, max_tokens: 64, cache_control: { type: "ephemeral" }, messages: user }),30 stop_sequences: (m) => ({ model: m, max_tokens: 64, stop_sequences: ["STOP"], messages: user }),31 web_search_2026: (m) => ({ model: m, max_tokens: 64, tools: [{ type: "web_search_20260209", name: "web_search", max_uses: 1 }], messages: user }),32 web_search_2025: (m) => ({ model: m, max_tokens: 64, tools: [{ type: "web_search_20250305", name: "web_search", max_uses: 1 }], messages: user }),33 code_execution_2026: (m) => ({ model: m, max_tokens: 64, tools: [{ type: "code_execution_20260521", name: "code_execution" }], messages: user }),34 max_tokens_over_cap: (m) => ({ model: m, max_tokens: 500000, messages: user }),35 speed_fast: (m) => ({ model: m, max_tokens: 64, speed: "fast", messages: user }),36 inference_geo_us: (m) => ({ model: m, max_tokens: 64, inference_geo: "us", messages: user }),37};3839