ComputeSage StackBench
GPU and LLM inference benchmarks, hardware evidence, deployment recommendations, and launch configs.
| Endpoint | https://mcp.computesage.com/mcp (streamable-http) |
|---|---|
| Registry version | 0.2.1 |
| Now | 5 tools · grade A (92) · pin ts_f282eba6f006efe9834ffa1ae46b683b · protocol 2025-06-18 |
| First / last listed | 2026-10-25 08:32 UTC / 2026-10-25 08:32 UTC |
| Grade over time | 2026-10-25: A (92), 5 tools |
Is this your server? Own a server? ToolFit hosted watch re-checks it every 24 h ($0.25 per 30 days) or 6 h ($0.60) and sends you an HMAC-signed webhook when its tools or grade change: POST https://api.youraitools.co/toolfit/v1/watch. To opt out, see About & opt-out.
Tools: search_evidence check_deployment_fit predict_performance recommend_deployment generate_launch_config
search_evidence | $0.00 free search of bounded public StackBench benchmark and failure evidence; use this for evidence-only requests. params: query*, limit |
check_deployment_fit | $0.01 USDC on Base mainnet: check model/configuration fit against bounded StackBench capacity evidence. params: model*, hardware_system*, quantization, engine, context_length*, concurrency, max_evidence |
predict_performance | $0.05 USDC: predict configuration throughput, memory, and power from bounded StackBench evidence with confidence, missing metrics, and caveats. params: model*, hardware_system*, quantization, engine, engine_version, kv_cache_type_k, kv_cache_type_v, prompt_tokens*, decode_context_depth, batch_size, ubatch_size, concurrency, flash_attention, requested_metrics, max_evidence |
recommend_deployment | $0.25 USDC: recommend what to buy or run from bounded StackBench fit, performance, confidence, evidence, caveats, and economics only when supported. params: model*, hardware_system, mode, quantization, engine, engine_version, prompt_tokens*, decode_context_depth, batch_size, ubatch_size, concurrency, flash_attention, candidate_kv_formats, minimum_prefill_tok_s, minimum_decode_tok_s, maximum_vram_bytes, maximum_power_watts, objective, evidence_policy, constraint_policy, recommendation_scope, quality_policy, quality_evidence_policy, workload_profile, workload_profile_version, quality_thresholds, quality_policy_version, max_results, budget_amount, budget_currency, maximum_tco_amount, electricity_rate_per_kwh, tco_years, utilization_hours_per_day, utilization_days_per_year, cost_policy, allow_manual_market_override, ownership_mode, manual_hardware_costs, economic_materiality_pct |
generate_launch_config | $0.50 USDC: generate deterministic vLLM or llama.cpp launch templates from an exact StackBench recommendation without claiming a launch occurred. params: runtime*, model*, hardware_system*, prompt_tokens*, quantization, engine_version, decode_context_depth, batch_size, ubatch_size, concurrency, flash_attention, candidate_kv_formats, minimum_prefill_tok_s, minimum_decode_tok_s, maximum_vram_bytes, maximum_power_watts, objective, constraint_policy, evidence_policy |
Entries describe changes in tool definitions. They are not statements about a server, its quality or its author. The grade measures how the tool definitions fit model providers (token cost, schema compatibility, confusable names) as scored by the ToolFit engine at the time of the listing. JSON: append .json. Opt-out: how or hello@youraitools.co.