8/18/2026, 8:14:28 AM · 15 tests · run 8be40889-2c96-4bda-9f25-5d9073fd2182
Plain streaming generation with a soft length constraint (~100 words).
{ "wordCount": 96, "outputTokens": 769 }
Asserts usage tokens populate and cost computes from the ChatReq contract.
{ "inputTokens": 64, "outputTokens": 243, "reasoningTokens": 0, "costUsd": 0.00037202, "paid": true }
Asserts a streaming call captures events that reconstruct into the final text.
{ "mode": "streaming", "streamEventCount": 178, "textBlocks": 1, "textLen": 35 }
Same prompt via stream() and complete() both return valid, matching results.
{ "streamText": "Paris", "completeText": "Paris" }
Asserts the model's reasoning knob behaves as its reasoning-config declares.
{ "class": "always-on", "reasoningTokens": 0, "outputTokens": 1492 }
Offers a calculator tool and checks the model calls it for an arithmetic question.
{ "toolCalled": true, "toolName": "multiply", "args": { "a": 23, "b": 17 }, "correct": true }
Forces a tool call, feeds the result back, and checks the continuation uses it.
{ "toolName": "multiply", "args": { "a": 23, "b": 17 }, "product": 391, "continuation": "23 multiplied by 17 is 391.", "usesResult": true, "forced": true }
Enables server-side web search and checks invocation counts, the per-request cost surcharge, and provider-executed output blocks.
{ "skipped": "no server-side web search support" }
Reasoning model: continuity token captured on a tool call and replayed so the follow-up succeeds.
{ "signatureType": "none", "tokenCaptured": false, "reasoned": false, "answered": true, "continuation": "Account **acme-42**\n* Current USD balance from `get_balance`\n\nbalanceUsd = 1234\n\nP = $1,234\nr = 5 % = 0.05 compounded annually\nn = 7 yr\n\nFut" }
Repeat a large prefix; assert cache-read tokens are reported and cost drops.
{ "skipped": "cacheStyle=none" }
Requests a strict JSON object and checks it parses with the required keys.
{ "parsed": { "name": "Anthropic", "founded": 2021, "headquarters": "San Francisco, California" }, "validJson": true, "missingKeys": [], "bareJson": true }
Summarizes an ~11k-token essay to ~500 words (long input, reasoning minimized).
{ "wordCount": 532, "inputTokens": 10220, "outputTokens": 980, "latencyMs": 10050 }
Tiny maxTokens on a long prompt yields finishReason:"length".
{ "finishReason": "length", "outputTokens": 100 }
Streaming-only models reject complete(); all others accept it.
{ "streamingOnly": false, "completeThrew": false }
A pre-aborted signal rejects the call (abort is threaded + honored).
{ "rejected": true, "errorName": "AbortError" }