Each row = tokens × its rate / 1,000,000. Cache reads are charged separately; reasoning is already included in output.
Astra: above 272K input tokens per request, input/cache rates double and output rates increase by 1.5×. ${number(s.longContext)} such requests in this selection. Claude Code uses its recorded list-price estimate, reconciled to the tiers above.${s.conditional?' Kinex + Opus on Vertex uses a 5-minute cache-write assumption; the recorded audit does not establish cache TTL.':''}${s.bounded?' A range allocates unclassified input across the possible rates; it does not treat missing cache counters as zero.':''}
API-equivalent estimates, not verified account charges. Subscription allocation, discounts, credits, taxes, simulator costs and archived attempts are excluded. Rates checked ${esc(UsageMetrics.pricingDate||'—')}.
${UsageMetrics.sources.map(x=>external(x.url,x.label)).join('')}Calculation data ↗
`;
}
function modelGroups() {
const groups=new Map();
for (const v of data.variants) {
const key=v.model+' / '+v.effort;
if(!groups.has(key))groups.set(key,{key,model:v.model,effort:v.effort,variants:[]});
groups.get(key).variants.push(v);
}
return [...groups.values()];
}
let expandedModels=null;
function comparisonCell(v,f) {
const s=stats(episodes(v.id,f.id)),link=`#runs/${v.id}?family=${f.id}`;
let value;
if(matrixMetric==='success')value=`
${matrixMetric==='tokens'?usageDefinition():matrixMetric==='cost'?'Estimated USD at the recorded or documented API rates. Expand a breakdown to inspect token categories and pricing tiers.':'Native passes / evaluated tasks. Pending tasks are excluded; compare coverage alongside success rates.'}
${modelGroups().map((g,i)=>`${esc(g.model)}${esc(g.effort)} effort · ${g.variants.map(v=>esc(v.agent)).join(' vs ')}${g.variants.length} harnesses ⌄
`;
}
function campaignProgress(){
const c=data.active_campaign;if(!c)return '';const n=c.counts||{};
return `BEHAVIOR evaluation · GPT-6 Astra / high
${number(n.finished||0)} / ${number(c.planned_episodes)} episodes completed · ${number(n.running||0)} running · ${number(n.queued||0)} queued${n.not_run?' · '+number(n.not_run)+' unavailable':''}${n.needs_review?' · '+number(n.needs_review)+' under review':''}. ${c.closed&&c.published_episodes===n.finished?'Evaluations have finished. All available results, videos and evidence are listed below.':'Results, videos and evidence are published after verification.'}
${c.quota_waiting?'Waiting for model-service quota recovery · ':''}${number(c.published_episodes)} episode records published · Results and protocol ↗
`;
}
function methodology() {
return `Reading the benchmarkProtocol, metrics & provenance+
One selected episode
Each task–agent pair shows its currently selected episode (seed 0 by default; explicit exceptions retain the measured seed). Previous attempts remain in the archive and are excluded from the displayed metrics. Results are added as episodes finish.
Native success
Success rate is passes / evaluated tasks, pooled over the selected episodes. Pending tasks do not count as failures. Results follow the original native verifier; Native notes identify verified simulator source issues.
Measured configuration
Each episode retains its actual instruction, control budget, runtime version and source revision. Historical runs retain their measured configuration. Availability is shown for each task and agent.
Token accounting
${usageDefinition()} Missing counters display as unavailable, never as zero.
${s.n?'Native success is computed from the selected evaluated episodes. Native notes identify simulator source issues. '+usageDefinition():'This combination has no published episodes. The catalog and planned budgets are shown below.'}
${taskBrowser(variant)}${methodology()}`;
renderRows();
}
function nativeNote(task,ep) {
const note=data.task_notes[task.key];if(!note)return '';
return ``;
}
function videoPanel(ep,mini=false) {
const m=ep.media||{},speed=m.speed||1,links=ep.links;
return `
Video unavailable. Open the pinned video directly using the Video link.
`;
}
function episodeMetrics(ep) {
const u=ep.usage||{},s=stats([ep]);
return `${usageTiles(s)}
${usageDefinition()}
${costBreakdown(s,true)}
${u.response_count==null&&Number.isFinite(u.known_response_count)?number(u.known_response_count)+'+':number(u.response_count)} model responses${number(ep.call_activity?.model_tool_calls)} tool calls${external(ep.links.usage,'Original usage record')}
`;
}
// Align words and punctuation while preserving the exact original whitespace.
function instructionDiff(original, revised) {
const split = text => text.match(/\s+|[\p{L}\p{N}_]+|[^\s]/gu) || [];
const a=split(original), b=split(revised);
const lcs=Array.from({length:a.length+1},()=>new Uint16Array(b.length+1));
for(let i=a.length-1;i>=0;i--)for(let j=b.length-1;j>=0;j--)
lcs[i][j]=a[i]===b[j]?lcs[i+1][j+1]+1:Math.max(lcs[i+1][j],lcs[i][j+1]);
const before=[],after=[];
const append=(list,text,changed)=>{const last=list.at(-1);if(last&&last.changed===changed)last.text+=text;else list.push({text,changed});};
let i=0,j=0;
while(i=lcs[i+1][j]))append(after,b[j++],true);
else append(before,a[i++],true);
}
const html=(parts,kind)=>parts.map(p=>p.changed&&p.text.trim()?`<${kind==='removed'?'del':'mark'} class="instruction-${kind}">${esc(p.text)}${kind==='removed'?'del':'mark'}>`:esc(p.text)).join('');
return {original:html(before,'removed'),revised:html(after,'added')};
}
function instructionPanel(task,ep) {
const native=ep?ep.native_instruction:task.native_instruction;
const measured=ep?(ep.instruction||(ep.instruction_policy==='original_native'?native:null)):task.catalog_instruction;
const changed=Boolean(native&&measured&&native!==measured);
const diff=changed?instructionDiff(native,measured):null;
const choices=[['measured',ep?'Measured':'Planned',measured],['native','Original native',native],['current','Current catalog',task.catalog_instruction]].filter(([key,,text])=>text&&(key!=='current'||(ep&&text!==measured)));
const currentChanged=native&&task.catalog_instruction&&native!==task.catalog_instruction;
return `
Task instruction
${ep?'The exact instruction used for this episode.':'The current catalog instruction for the planned evaluation.'}
${choices.map(([k,label],i)=>``).join('')}
${choices.map(([k,,text],i)=>{
const comparison=k==='current'?(currentChanged?instructionDiff(native,text):null):diff;
const html=comparison?(k==='native'?comparison.original:comparison.revised):esc(text);
const note=comparison?(k==='native'?'Struck-through text was removed or replaced in the '+(ep?'measured':'planned')+' instruction.':'Highlighted text was added or replaced relative to the original native instruction.'):(native?'No changes from the original native instruction.':'Original native text is unavailable for comparison.');
return `
${comparison?``:''}${note}${k==='current'?' This catalog text is not the measured episode instruction.':''}
${html}
`;
}).join('')}`;
}
function overviewDetail(task,ep) {
const initializationBlocked=(data.active_campaign?.initialization_blocked||[]).includes(task.key+'/'+current.variant);
if(!ep)return `
${initializationBlocked?'No model was called. Seed 0 is retained by user decision; this is excluded from task success rates.':'This task is included in the planned catalog. Results and usage will appear after evaluation.'}
${m.source_fps?m.source_fps+' FPS source capture · ':''}${m.speed||1}× simulation speed. Session timestamps use wall time and are not synchronized to video frames.
${nativeNote(task,ep)}${episodeMetrics(ep)}
${instructionPanel(task,ep)}
Episode notes
Recorded observations and task-specific interpretation.
${reason?`
${esc(reason)}
`:`
${esc(a.outcome||'The native result and original evidence are retained for this episode.')}
`}Read recorded analysis${Object.entries(a).filter(([k])=>k!=='preflight').map(([k,v])=>`
No standalone tool or memo files were saved in this episode. The original session and workspace evidence remain available.
'}
${evidenceCard('Execution evidence','Original verdict, protocol and runtime identity.',[['verdict','Native verdict'],['protocol','Protocol'],['provenance','Source identity']],ep)}${evidenceCard('Session & interactions','Original tool calls, observations and edited workspace.',[['native_session','Native session'],['trace','Trace'],['workspace_changes','Workspace changes']],ep)}
`;
}
async function fetchText(url) {
if(fileCache.has(url))return fileCache.get(url);
const r=await fetch(url);if(!r.ok)throw Error('HTTP '+r.status);
if(Number(r.headers.get('content-length'))>3e6)throw Error('File exceeds the 3 MB preview limit; use Open file');
const text=await r.text();if(text.length>3e6)throw Error('File exceeds the preview limit; use Open file');
fileCache.set(url,text);if(fileCache.size>20)fileCache.delete(fileCache.keys().next().value);return text;
}
async function loadResource(index,token) {
const r=current.ep?.resources?.[index];if(!r)return;
current.resource=index;$('#resource-name').textContent=r.name;$('#resource-open').href=r.file;$('#resource-kind').textContent=r.kind||'Final workspace snapshot';$('#resource-preview').textContent='Loading file…';
$$('[data-resource]').forEach(b=>{b.classList.toggle('active',+b.dataset.resource===index);b.setAttribute('aria-pressed',+b.dataset.resource===index);});
try{const text=await fetchText(r.file);if(token===epoch&¤t.resource===index){$('#resource-preview').textContent=text;current.resourceText=text;}}
catch(e){if(token===epoch&¤t.resource===index){$('#resource-preview').textContent='Preview unavailable: '+e.message;current.resourceText=null;}}
}
const evidenceGroups=[
['Native evaluation','Owner verdicts, measured task and control protocol.',[['verdict','Native verdict'],['protocol','Measured protocol'],['native_goal','Task instructions'],['owner_journal','Owner action journal']]],
['Session & workspace','The original agent trajectory and final files.',[['native_session','Native session'],['transcript','Readable transcript'],['trace','Agent trace'],['workspace','Workspace archive'],['workspace_changes','Workspace manifest']]],
['Recording & review','Published media and checks against the recorded evidence.',[['video','Video'],['poster','Poster'],['terminal_frame','Terminal frame'],['recording_manifest','Recording manifest'],['media_check','Media validation'],['analysis','Episode analysis']]],
['Usage & provenance','Provider counters and the actual code used by the episode.',[['usage','Token usage'],['provider_usage','Provider usage ledger'],['provenance','Source identities'],['tool_errors','Tool errors']]]
];
function evidenceCard(title,description,items,ep) {
return `