badgr-cli 1.0.19 → 1.0.21

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "badgr-cli",
-  "version": "1.0.19",
+  "version": "1.0.21",
   "description": "Badgr, run or serve GPU workloads from one command",
   "type": "module",
   "bin": {

package/src/badgr.js

@@ -26,10 +26,14 @@ ${chalk.bold('COMMANDS')}
   ${chalk.cyan('badgr capacity')}               Check what GPU capacity is available right now
 
 ${chalk.bold('EXAMPLES')}
-  ${chalk.dim('# Simplest — Badgr picks the GPU:')}
+  ${chalk.dim('# Simplest — Badgr picks the GPU (RunPod, reliable):')}
   badgr run python train.py
   badgr serve meta-llama/Llama-3.1-8B-Instruct
 
+  ${chalk.dim('# Opt into cheaper budget providers (Vast.ai etc.):')}
+  badgr run python train.py --cheap
+  badgr serve meta-llama/Llama-3.1-8B-Instruct --cheap
+
   ${chalk.dim('# Pin a specific GPU:')}
   badgr run python train.py --gpu A100
   badgr serve meta-llama/Llama-3.1-8B-Instruct --gpu L40S
@@ -44,10 +48,12 @@ ${chalk.bold('EXAMPLES')}
   badgr receipts dep-abc123
 
 ${chalk.bold('badgr run OPTIONS')}
-  --gpu <type>           GPU type (default: auto — Badgr picks cheapest available)
+  --gpu <type>           GPU type (default: auto — Badgr picks best available on RunPod)
+  --cheap                Search budget providers too (Vast.ai etc.) for lower prices
   --image <image>        Docker image (default: python:3.11-slim)
   --count <n>            Number of GPUs (default: 1)
   --region US|EU|AU      Region preference
+  --tier 1|2             Provider tier: 1 = reliable (default), 2 = budget
   --max-price <$/hr>     Hard spend cap per GPU-hour
   --max-runtime <min>    Auto-stop after N minutes (recommended)
   --max-cost <$>         Auto-stop when spend reaches this amount
@@ -55,8 +61,10 @@ ${chalk.bold('badgr run OPTIONS')}
 
 ${chalk.bold('badgr serve OPTIONS')}
   --gpu <type>           GPU type (default: auto — inferred from model size)
+  --cheap                Search budget providers too (Vast.ai etc.) for lower prices
   --count <n>            Number of GPUs (default: 1)
   --region US|EU|AU      Region preference
+  --tier 1|2             Provider tier: 1 = reliable (default), 2 = budget
   --max-price <$/hr>     Hard spend cap per GPU-hour
   --no-wait              Skip endpoint health check
 

package/src/commands/run.js

@@ -18,6 +18,8 @@ export function parseRunArgs(args) {
     if (args[i] === '--image')          { flags.image = args[++i]; i++; continue; }
     if (args[i] === '--count')          { flags.count = parseInt(args[++i], 10); i++; continue; }
     if (args[i] === '--region')         { flags.region = args[++i]; i++; continue; }
+    if (args[i] === '--tier')           { flags.tier = args[++i]; i++; continue; }
+    if (args[i] === '--cheap')          { flags.cheap = true; i++; continue; }
     if (args[i] === '--max-price')      { flags.maxPrice = parseFloat(args[++i]); i++; continue; }
     if (args[i] === '--name')           { flags.name = args[++i]; i++; continue; }
     if (args[i] === '--detach')          { flags.detach = true; i++; continue; }
@@ -268,11 +270,12 @@ async function attachToJob(config, depId, { chalk, maxRuntimeMs = null, maxCost
   }
 }
 
-// Ask the backend for the cheapest available GPU right now.
+// Ask the backend for the best available GPU right now.
+// Defaults to tier 1 (RunPod) for reliability; pass tier='2' for budget providers.
 // Returns { gpu, region, price } or null when nothing is available.
-async function findAutoGpu(config, chalk) {
+async function findAutoGpu(config, chalk, tier = '1') {
   try {
-    const params = new URLSearchParams({ max_price: '10' });
+    const params = new URLSearchParams({ max_price: '10', tier });
     return await callApi(`/capacity/auto?${params}`, {
       apiKey: config.apiKey,
       baseUrl: config.baseUrl,
@@ -306,6 +309,12 @@ export async function runCommand(config, args, chalk) {
   const maxRuntimeMs = flags.maxRuntime ? flags.maxRuntime * 60 * 1000 : null;
   const maxCost      = flags.maxCost ?? null;
 
+  // Resolve effective tier: --cheap and --tier 2 opt into budget providers;
+  // everything else defaults to tier 1 (RunPod only) for reliability.
+  const effectiveTier = (flags.cheap || flags.tier === '2' || flags.tier === 'tier2' || flags.tier === 'tier-2')
+    ? '2'
+    : (flags.tier || '1');
+
   // ── Auto GPU selection (no --gpu specified) ────────────────────────────────
   let gpu = flags.gpu ? flags.gpu.toUpperCase().replace('-', '_') : null;
   let autoRegion = null;
@@ -314,12 +323,13 @@ export async function runCommand(config, args, chalk) {
     console.log(chalk.bold('\n⚡ Running GPU job\n'));
     if (command) console.log(`  ${chalk.bold('Command:')}  ${command.join(' ')}`);
     if (image)   console.log(`  ${chalk.bold('Image:')}    ${image}`);
+    if (effectiveTier === '2') console.log(`  ${chalk.dim('(budget mode — searching all providers)')}`);
     console.log();
     process.stdout.write(chalk.dim('  Finding GPU...'));
 
     let best;
     try {
-      best = await findAutoGpu(config, chalk);
+      best = await findAutoGpu(config, chalk, effectiveTier);
     } catch (err) {
       process.stdout.write('\n');
       console.error(chalk.red(`\n  ✗ Could not find GPU capacity: ${err.message}\n`));
@@ -384,6 +394,7 @@ export async function runCommand(config, args, chalk) {
       ...(effectiveRegion ? { region: effectiveRegion } : {}),
       max_price_per_hour: flags.maxPrice,
       name: flags.name,
+      tier: effectiveTier,
     };
   }
 
@@ -436,7 +447,7 @@ export async function runCommand(config, args, chalk) {
         const d2 = err2.errorData;
         if (d2?.code === 'PROVISIONING_FAILED' || d2?.code === 'PROVIDER_ADAPTER_ERROR') {
           if (d2?.low_cost_provider_failed) {
-            console.error(chalk.red(`\n  ✗ Low-cost provider failed. Primary provider also unavailable.\n`));
+            console.error(chalk.red(`\n  ✗ Tier 2 unavailable. Tier 1 also unavailable. Try again shortly.\n`));
           } else {
             console.error(chalk.red(`\n  ✗ Badgr found ${chosen.gpu} capacity but could not start the machine. Please try again.\n`));
           }
@@ -453,7 +464,7 @@ export async function runCommand(config, args, chalk) {
       }
     } else if (d?.code === 'PROVISIONING_FAILED' || d?.code === 'PROVIDER_ADAPTER_ERROR') {
       if (d?.low_cost_provider_failed) {
-        console.error(chalk.red(`\n  ✗ Low-cost provider failed. Primary provider also unavailable.\n`));
+        console.error(chalk.red(`\n  ✗ Tier 2 unavailable. Tier 1 also unavailable. Try again shortly.\n`));
       } else {
         console.error(chalk.red(`\n  ✗ Badgr found ${gpu} capacity but could not start the machine. Please try again.\n`));
       }
@@ -471,8 +482,8 @@ export async function runCommand(config, args, chalk) {
     }
   }
 
-  if (dep.provider_fallback_note === 'low_cost_provider_failed') {
-    console.log(chalk.yellow('  ℹ Low-cost provider failed. Using primary provider instead.\n'));
+  if (dep.provider_fallback_note === 'tier2_failed_using_tier1') {
+    console.log(chalk.yellow('  ℹ Tier 2 unavailable. Running on Tier 1 instead.\n'));
   }
 
   const rcptId = dep.receipt_id || generateReceiptId();
@@ -488,6 +499,7 @@ export async function runCommand(config, args, chalk) {
   console.log();
   console.log(chalk.bold(`  Job ID:    ${chalk.cyan(dep.deployment_id)}`));
   console.log(`  ${chalk.bold('GPU:')}      ${dep.gpu_type} × ${dep.gpu_count}`);
+  if (dep.tier) console.log(`  ${chalk.bold('Tier:')}     ${dep.tier}`);
   if (dep.cost_per_hour > 0) console.log(`  ${chalk.bold('Rate:')}     $${dep.cost_per_hour.toFixed(2)}/hr`);
   console.log(`  ${chalk.bold('Receipt:')}  ${chalk.dim(rcptId)}`);
 

package/src/commands/serve.js

@@ -17,6 +17,8 @@ export function parseServeArgs(args) {
     if (args[i] === '--gpu')        { flags.gpu = args[++i]; i++; continue; }
     if (args[i] === '--count')      { flags.count = parseInt(args[++i], 10); i++; continue; }
     if (args[i] === '--region')     { flags.region = args[++i]; i++; continue; }
+    if (args[i] === '--tier')       { flags.tier = args[++i]; i++; continue; }
+    if (args[i] === '--cheap')      { flags.cheap = true; i++; continue; }
     if (args[i] === '--max-price')  { flags.maxPrice = parseFloat(args[++i]); i++; continue; }
     if (args[i] === '--name')       { flags.name = args[++i]; i++; continue; }
     if (args[i] === '--no-wait')    { flags.noWait = true; i++; continue; }
@@ -63,9 +65,15 @@ export async function serveCommand(config, args, chalk) {
   const gpu = flags.gpu ? flags.gpu.toUpperCase().replace('-', '_') : 'AUTO';
   const gpuLabel = gpu === 'AUTO' ? 'auto' : gpu;
 
+  // Default to tier 1 (RunPod) for reliability; --cheap or --tier 2 opts into budget providers.
+  const effectiveTier = (flags.cheap || flags.tier === '2' || flags.tier === 'tier2' || flags.tier === 'tier-2')
+    ? '2'
+    : (flags.tier || '1');
+
   console.log(chalk.bold('\nServing model\n'));
   console.log(`  ${chalk.bold('Model:')}  ${model}`);
   console.log(`  ${chalk.bold('GPU:')}    ${gpuLabel}`);
+  if (effectiveTier === '2') console.log(`  ${chalk.dim('(budget mode — searching all providers)')}`);
   console.log();
   process.stdout.write(chalk.dim('  Finding GPU capacity...\n'));
 
@@ -79,6 +87,7 @@ export async function serveCommand(config, args, chalk) {
       ...(regionOverride || effectiveRegion ? { region: regionOverride || effectiveRegion } : {}),
       max_price_per_hour: flags.maxPrice,
       name: flags.name,
+      tier: effectiveTier,
     };
   }
 
@@ -125,7 +134,7 @@ export async function serveCommand(config, args, chalk) {
         const d2 = err2.errorData;
         if (d2?.code === 'PROVISIONING_FAILED' || d2?.code === 'PROVIDER_ADAPTER_ERROR') {
           if (d2?.low_cost_provider_failed) {
-            console.error(chalk.red(`\n  ✗ Low-cost provider failed. Primary provider also unavailable.\n`));
+            console.error(chalk.red(`\n  ✗ Tier 2 unavailable. Tier 1 also unavailable. Try again shortly.\n`));
           } else {
             console.error(chalk.red(`\n  ✗ Badgr found ${chosen.gpu} capacity but could not start the endpoint. Please try again.\n`));
           }
@@ -137,7 +146,7 @@ export async function serveCommand(config, args, chalk) {
       }
     } else if (d?.code === 'PROVISIONING_FAILED' || d?.code === 'PROVIDER_ADAPTER_ERROR') {
       if (d?.low_cost_provider_failed) {
-        console.error(chalk.red(`\n  ✗ Low-cost provider failed. Primary provider also unavailable.\n`));
+        console.error(chalk.red(`\n  ✗ Tier 2 unavailable. Tier 1 also unavailable. Try again shortly.\n`));
       } else {
         console.error(chalk.red(`\n  ✗ Badgr found capacity but could not start the endpoint. Please try again.\n`));
       }
@@ -161,8 +170,8 @@ export async function serveCommand(config, args, chalk) {
     }
   }
 
-  if (dep.provider_fallback_note === 'low_cost_provider_failed') {
-    console.log(chalk.yellow('  ℹ Low-cost provider failed. Using primary provider instead.\n'));
+  if (dep.provider_fallback_note === 'tier2_failed_using_tier1') {
+    console.log(chalk.yellow('  ℹ Tier 2 unavailable. Running on Tier 1 instead.\n'));
   }
 
   addDeployment({
@@ -215,6 +224,7 @@ export async function serveCommand(config, args, chalk) {
   console.log(`  ${chalk.bold('Base URL:')}   ${chalk.cyan(endpointUrl)}`);
   console.log(`  ${chalk.bold('Model:')}      ${dep.model || model}`);
   console.log(`  ${chalk.bold('GPU:')}        ${dep.gpu_type} × ${dep.gpu_count}`);
+  if (dep.tier) console.log(`  ${chalk.bold('Tier:')}       ${dep.tier}`);
   if (dep.cost_per_hour > 0) console.log(`  ${chalk.bold('Rate:')}       $${dep.cost_per_hour.toFixed(2)}/hr`);
   console.log(`  ${chalk.bold('Logs:')}       ${chalk.dim(`badgr logs ${dep.deployment_id}`)}`);
   console.log(`  ${chalk.bold('Stop billing:')} ${chalk.dim(`badgr down ${dep.deployment_id}`)}`);

package/tests/commands.test.js

@@ -71,6 +71,21 @@ describe('parseRunArgs', () => {
     const { flags } = parseRunArgs(['python', 'train.py', '--max-cost', '5.00']);
     expect(flags.maxCost).toBe(5.0);
   });
+
+  it('parses --cheap flag', () => {
+    const { flags } = parseRunArgs(['python', 'train.py', '--cheap']);
+    expect(flags.cheap).toBe(true);
+  });
+
+  it('--cheap defaults to falsy when not passed', () => {
+    const { flags } = parseRunArgs(['python', 'train.py']);
+    expect(flags.cheap).toBeFalsy();
+  });
+
+  it('parses --tier 2 flag', () => {
+    const { flags } = parseRunArgs(['python', 'train.py', '--tier', '2']);
+    expect(flags.tier).toBe('2');
+  });
 });
 
 describe('classifyFailure', () => {
@@ -187,6 +202,16 @@ describe('parseServeArgs', () => {
     const { model } = parseServeArgs(['--gpu', 'RTX_4090']);
     expect(model).toBeNull();
   });
+
+  it('parses --cheap flag', () => {
+    const { flags } = parseServeArgs(['my/model', '--cheap']);
+    expect(flags.cheap).toBe(true);
+  });
+
+  it('--cheap defaults to falsy when not passed', () => {
+    const { flags } = parseServeArgs(['my/model']);
+    expect(flags.cheap).toBeFalsy();
+  });
 });
 
 describe('promptFallback output', () => {