Reasoning And Thinking
xAI API (Grok) · xAI
Foundation Model & API — xAI (Grok)
Evaluates xAI's Reasoning & Thinking across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
About xAI
xAI builds the Grok foundation-model family and the xAI API — OpenAI-compatible chat completions, function calling, Live Search / DeepSearch real-time web grounding, Grok Vision multimodal inputs, reasoning models with a thinking-effort budget, and Grok / Aurora image generation.
Sample tests· showing 3 of 9
| # | Input | Expected behavior | Check |
|---|---|---|---|
| 01 | Agent enables reasoning by setting reasoning_effort='high' (or 'low') on a Grok reasoning-capable model for a multi-step problem. | Pin the model id to a reasoning-capable Grok variant. Set reasoning_effort to one of the documented enum values (low / medium / high) per docs.x.ai. [REQUIRES-VERIFICATION] on exact field name and supported values. Trade higher latency for deeper reasoning; reserve 'high' for complex tasks. | Pass / FailAi Platformhigh |
| 02 | User-facing UI has a 30s timeout. Agent enables reasoning_effort='high' and observes p95 latency exceeding the SLA. | Higher reasoning effort raises latency. Set client read timeouts ABOVE the expected p99 (e.g., 60s+ for high effort) [REQUIRES-VERIFICATION] for current latency profiles. For low-latency UI, prefer reasoning_effort='low' or a non-reasoning model and stream tokens to mask wait time. | Pass / FailAi Platformhigh |
| 03 | Reasoning models bill thinking tokens separately from visible output tokens per docs.x.ai pricing [REQUIRES-VERIFICATION]. Finance reports a cost spike disproportionate to visible answer length. | Read usage.completion_tokens_details.reasoning_tokens (OpenAI-compat field) — bill at the documented rate. Treat reasoning tokens as a separate line item from visible completion in cost telemetry. | Pass / FailAi Platformhigh |
How this eval is graded
Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
Rubric criteria
- Xai
- Ai Platform
- Reasoning And Thinking
Recommended for
Works with
Related evals
Claude API
Evaluates Anthropic's Batch API across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Extended Thinking across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Files API & Citations across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
ViewFrequently asked questions
What does the Reasoning And Thinking eval for xAI xAI API (Grok) test?+
Evaluates xAI's Reasoning & Thinking across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
How is the Reasoning And Thinking eval scored?+
The judge rubric: Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
How many test cases does this eval pack include?+
The Reasoning And Thinking pack for xAI xAI API (Grok) contains 9 test cases. 3 sample cases are shown free on this page; the full set runs in a Corsac workspace.
How do I run this eval?+
Sign up for Corsac, connect your model or agent endpoint, and run the Reasoning And Thinking pack as-is or after customizing thresholds. Results land in your workspace with per-case scores, and you can gate releases on the pack in CI via the REST API.
Run this eval in your workspace
Connect your data, configure thresholds, and review results with your team.