Blaxel

45/100

Serverless sandboxes for AI agents to run untrusted code.

blaxel.ai

// scorecard · 45/100

verified 1mo ago

methodology →

Agent-readiness23/35

Agent registration (api_key)6/9

Public API5/5

OpenAPI spec0/9

MCP server9/9

llms.txt3/3

Reliability & performance0/15

Actively maintained—

Production-ready0/4

Updated recently—

Pricing6/13

Itemized public pricing6/6

Free tier / trial0/4

Free-tier generosity—

Docs & DX2/12

Docs quality—

Public API2/2

Quickstart / examples0/2

Security & compliance10/12

SOC 26/6

ISO 270012/2

GDPR0/2

Security / trust page2/2

Openness4/10

Open source0/6

Open / un-gated API4/4

Company maturity0/3

Company size—

Years operating—

// rankings · cost per standard workload

#9 in Code Sandboxes$82.8/ 1k sandbox-hrs

Memory-only $0.0000115/GB-s x2GB = $0.000023/s x3.6M = $82.80/mo (CPU incl.)

blaxel.ai/pricing

// overview

Blaxel provides infrastructure for autonomous agents with compute primitives (sandboxes as microVMs, batch jobs, MCP servers), storage (Agent Drive distributed filesystem, volumes, local FS snapshots), and networking (outbound controls, static IPs, proxy routing, model gateway). Sandboxes boot in milliseconds or resume in ~25ms with memory and filesystem state intact, supporting up to 50K+ concurrent instances that idle to zero cost. It is offered as managed cloud or self-hosted VPC with fully usage-based pricing and no base subscription.

Best for: Teams building the next generation of autonomous agents

// pricing

Pay-as-you-go$0 + usage

Up to $200 free credits
Access to all products
Tier system for higher limits

CustomContact for quote

All pay-as-you-go features
Dedicated deployments
Up to 256 GB RAM

// security & compliance

✓ SOC 2✓ ISO 27001✗ GDPR✗ HIPAA

// features

MicroVM sandboxes boot in milliseconds and resume in ~25ms with state intact
Agent Drive distributed filesystem for multi-agent collaboration
Batch jobs spawn thousands of tasks across sandboxes in seconds
MCP servers run as first-class private workloads
Persistent environments idle to zero with $0 compute cost while suspended
Model Gateway as one co-located endpoint for every model provider