At a glance — Oct 2026
Updated 2026-10-27 · forward this page: /jtbds/agent-backend#briefScores as of 2026-10-19 (8 days old).
- Depot Pick
- Modal · 8.44/10 for this job (curated estimate, high confidence)
capability for long jobs + dx + reliability.
- Runner-up
- Convex · 8.15/10
Scores higher than the pick on: Maturity / track record.
- When not to use the pick
- Already on K8s/EC2 with good orchestration — stay. Pure edge short tools — CF Workers.
- Lock-in / migration
- The pick has lower stored lock-in risk than runner-up Convex (4.5 vs 6.2).
- Sourced price (no comparable unit)
Pick: no sourced price in this job's unit (no_matching_sourced_price).
No cheapest SKU (no_comparable_unit_for_job). No price was invented.
- Free option
- No free resource lists this job. A cheap paid SKU is not a free option.
- Failure modes to watch
- Instruction drift: Restate the stop condition in the trace. End the run when it is met. Review long traces for steps that do not serve the original task.
- Inter-agent misalignment: Log the assigned subtask next to the worker's stated result. The lead rejects a result that does not match the assignment.
- Runaway spend: Set a budget and a tool-call cap per task. Alert when spend jumps. Stop the run at the cap instead of summarizing a partial trace as success.
Cite: Agent Provider Depot, Depot Pick for agent-backend: modal, scores as of 2026-10-19. Agents: GET /api/decide?jtbd=agent-backend
Modal
capability for long jobs + dx + reliability.
Pricing (directional): As of 2026-10-24 (modal.com/pricing, fetched 2026-10-24): Python-native serverless GPU/CPU billed per second of compute ('You never pay for idle resources'). GPU list rates per second: H100 SXM5 '$0.001097', H200 SXM '$0.001261', B200 '$0.0…
Lock-in risk: higher = more lock-in (worse). Overall already inverts this dimension.
Depot darling for Python agent backends and bursty GPU jobs. If you think in functions not VMs, start here.
Quote for Long-running agent worker backend
Depot Pick stays Modal. Prices below are sourced SKU rows only (as of 2026-10-27). Input and output are not blended. GET /api/quote?jtbd=agent-backend
Nothing was compared. no comparable sourced unit for this job; chat input $/1M and other units were not borrowed pickIsPriceWinner is null.
- CPU physical core usd_per_core_second $0.0000131 source as of 2026-10-27
- Memory usd_per_gib_second $0.00000222 source as of 2026-10-27
- Network egress usd_per_gib $0.04 source as of 2026-10-27
- Nvidia A10 usd_per_gpu_second $0.000306 source as of 2026-10-27
- Nvidia A100 40GB usd_per_gpu_second $0.000583 source as of 2026-10-27
- Nvidia A100 80GB usd_per_gpu_second $0.000694 source as of 2026-10-27
- Nvidia B200 usd_per_gpu_second $0.001736 source as of 2026-10-27
- Nvidia B300 usd_per_gpu_second $0.001972 source as of 2026-10-27
- Nvidia H100 SXM5 usd_per_gpu_second $0.001097 source as of 2026-10-27
- Nvidia H200 SXM usd_per_gpu_second $0.001261 source as of 2026-10-27
- Nvidia L4 usd_per_gpu_second $0.000222 source as of 2026-10-27
- Nvidia L40S usd_per_gpu_second $0.000542 source as of 2026-10-27
- Nvidia RTX PRO 6000 usd_per_gpu_second $0.000842 source as of 2026-10-27
- Nvidia T4 usd_per_gpu_second $0.000164 source as of 2026-10-27
- Sandbox CPU usd_per_core_second $0.00003942 source as of 2026-10-27
- Sandbox memory usd_per_gib_second $0.00000667 source as of 2026-10-27
- Team plan usd_per_month $250 source as of 2026-10-27
- Volumes usd_per_gib_month $0.09 source as of 2026-10-27
11 of 16 candidates have no sourced price row. 16 are excluded from the Matching price comparison (none). no comparable sourced unit for this job; chat input $/1M and other units were not borrowed
Runners-up
Convex
Great TS agent product backend when you want reactive state+actions together. Prefer Supabase/Neon when…
Fly.io
Excellent for globally distributed agent sidecars and Machines-based sandboxes. Prefer Railway/Render if you…
Railway
Depot favorite for shipping agent APIs quickly. When bills climb, move steady-state to Fly/Hetzner/DO.
Render
Boring reliable PaaS. Great when Railway feels too hip and AWS feels too heavy.
AWS EC2
Default enterprise VM substrate. Agents should wrap it — humans shouldn't live in the console.
When NOT to use the top pick
Already on K8s/EC2 with good orchestration — stay. Pure edge short tools — CF Workers.
Full ranking (score table)
Ranked with this JTBD's weights. Scores are curated estimates unless marked otherwise. Agents: prefer /api/decide then this table for tie-breaks.
| # | Provider | Overall | Fit | $/perf | Latency | DX | As of |
|---|---|---|---|---|---|---|---|
| 1 | Cloudflare Workers | 8.5 | 8.6 | 9.2 | 9.2 | 8.5 | 2026-09-14 |
| 2 | ModalDepot Pick | 8.4 | 9.0 | 8.2 | 8.6 | 9.6 | 2026-10-19 |
| 3 | Upstash QStash | 8.4 | 8.2 | 9.2 | 8.4 | 9.2 | 2026-09-14 |
| 4 | Fly.io | 8.3 | 8.8 | 8.2 | 8.0 | 8.8 | 2026-09-14 |
| 5 | Convex | 8.2 | 8.6 | 8.0 | 8.6 | 9.2 | 2026-09-14 |
| 6 | Deno Deploy | 8.1 | 7.8 | 8.5 | 8.6 | 9.0 | 2026-09-14 |
| 7 | Render | 8.1 | 8.0 | 7.5 | 8.8 | 9.0 | 2026-09-14 |
| 8 | Railway | 8.1 | 8.2 | 7.8 | 8.4 | 9.4 | 2026-09-14 |
| 9 | Inngest | 8.1 | 8.2 | 8.0 | 8.4 | 9.0 | 2026-09-14 |
| 10 | Temporal Cloud | 8.0 | 9.0 | 6.8 | 9.2 | 7.2 | 2026-09-14 |
| 11 | AWS EC2 | 8.0 | 9.2 | 7.0 | 9.5 | 6.5 | 2026-09-14 |
| 12 | Trigger.dev | 7.8 | 7.8 | 8.2 | 8.0 | 9.2 | 2026-09-14 |
| 13 | Northflank | 7.5 | 8.0 | 7.4 | 7.8 | 8.0 | 2026-09-14 |
| 14 | LangGraph Platform / LangSmith Deploy | 7.5 | 8.4 | 6.8 | 8.0 | 8.2 | 2026-09-14 |
| 15 | Anyscale | 7.5 | 8.2 | 6.5 | 8.0 | 7.5 | 2026-09-14 |
| 16 | Letta Cloud | 7.4 | 8.0 | 7.8 | 7.2 | 8.0 | 2026-09-14 |
Score methodology for this job
capability for long jobs + dx + reliability.
capabilityFit: 25%pricePerformance: 18%latencyReliability: 15%developerDx: 15%privacyPosture: 10%lockInRisk: 7%maturity: 10%