predict performance

$0.05 USDC: predict configuration throughput, memory, and power from bounded StackBench evidence with confidence, missing metrics, and caveats.

liveverifiedunknown~5s typical

mcpprobedcomputesagestackbench

$0.0020

per call

plus $0.001 platform fee per run

Open MCP server; Cracked bills routing only

Provider Upstream docs

$Set up https://cracked-ai-pearl.vercel.app/SKILL.md, then use Cracked to run mcp-com-computesage-stackbench /predict_performance for me.

Input

Field	Type	Description
model*	string
hardware_system*	string
quantization	string
engine	string
engine_version	string
kv_cache_type_k	string
kv_cache_type_v	string
prompt_tokens*	integer
decode_context_depth	string
batch_size	string
ubatch_size	string
concurrency	integer	default 1
flash_attention	string
requested_metrics	array
max_evidence	integer	default 12

Call it

curl

curl https://cracked-ai-pearl.vercel.app/v1/run \
  -H "Authorization: Bearer ck_live_..." -H "content-type: application/json" \
  -d '{"provider":"mcp-com-computesage-stackbench","endpoint":"/predict_performance","input":{"model":"...","hardware_system":"...","prompt_tokens":1,"concurrency":1,"max_evidence":12}}'

cli

npx cracked-ai run -p mcp-com-computesage-stackbench -e /predict_performance -i '{"model":"...","hardware_system":"...","prompt_tokens":1,"concurrency":1,"max_evidence":12}'

mcp

run_tool({ provider: "mcp-com-computesage-stackbench", endpoint: "/predict_performance", input: {"model":"...","hardware_system":"...","prompt_tokens":1,"concurrency":1,"max_evidence":12} })

Try it

Runs against your signed-in workspace balance. Sign in if you have not.