open source · self-host · your keys, 0% markup

The flight recorder for AI agents

Every call, traced. Every trace, provable.

Start free GitHub no credit card · apache-2.0 core
sha-256 hash chain ed25519-signed merkle root 191 providers one rust binary

Hover a step to see it recorded.

tracelane — replay

191 providers, one canonical wire format

OpenAI
Anthropic
Google
Mistral
Llama
Bedrock
Groq
191 providers

The problem

Agents fail without crashing.

State drift compounds
A 200 doesn't mean it worked
Every tool call is code running on your behalf

How it works

Record. Replay. Prove.

One Rust binary. Open core, Apache 2.0.

Record

Every call, full fidelity, your keys.

Replay

Step to the exact call that broke.

Prove

Hash-chained, signed, verifiable offline.

Measured, not claimed

2ms added at p50. 4ms at p95. 5ms at p99.

Gateway overhead on top of the model call, with capture on — measured with LiteLLM's own open AIGatewayBench harness, not one we wrote ourselves.

Gateway overhead in milliseconds added over a direct call to the mock upstream, run 2 (run 1 in parentheses where it differs)
GatewayVersionp50p95p99Peak RSS
Tracelane 1f3af9ad 2 ms 4 ms 5 ms 14 MB
Bifrost v2.0.0 2 ms 2 ms 3 ms 101 MB
Portkey (OSS) 1.15.2 3 ms (run 1: 4 ms) 5 ms (run 1: 7 ms) 8 ms (run 1: 9 ms) 177 MB (run 1; run 2 not sampled)
LiteLLM Python 1.100.0 14 ms (run 1: 15 ms) 34 ms (run 1: 36 ms) 47 ms (run 1: 52 ms) 323 MB
LiteLLM Rust not measured not measured not measured not measured not measured

Peak memory (RSS), gateway process only

Tracelane 14 MBBifrost 101 MBPortkey (OSS) 177 MB (run 1; run 2 not sampled)LiteLLM Python 323 MB

Direct-to-mock baseline (already subtracted from every figure above): 1 / 2 / 3 ms p50/p95/p99. Host: one Hetzner CCX23 (4 vCPU, 16 GB). 16 concurrent clients, 30 seconds, two runs.
† Bifrost via npx -y @maximhq/bifrost, both runs — its version string is only emitted on first install, so run 2's CSV records it as "unknown"; v2.0.0 is from run 1.
‡ LiteLLM Rust: as of 2026-09-07 no public build exists — the harness's own notes build it from a private branch on LiteLLM's machine. We'll run it when one is published.

bash bench/gateway/aigatewaybench/run.sh Full method, caveats and the raw CSVs →

How we compare

Thirteen claims, sourced.

Against each vendor's own docs and pricing pages, fetched 2026-09-07. A dash means we didn't find it documented, not that the vendor lacks it.

Feature comparison: Tracelane, LiteLLM, Portkey, Langfuse, LangSmith
Tracelane LiteLLM Portkey Langfuse LangSmith
Measured gateway overhead (p50) 2 ms † Python proxy 14–15 ms, same bench; Rust build not public 3–4 ms, same bench run no gateway no gateway
Rust hot path Python proxy measured; Rust gateway announced, no public build TypeScript/Node.js no gateway no gateway
BYOK, 0% token markup lists an LLM Gateway; BYOK not documented
OTel-native ingest partial partial partial
Multi-agent swimlanes
Sessions per-user identity on the roadmap
Eval loop — datasets, experiments, online evals, annotation queues, CI gate proven on prod
Prompt management + playground
MCP server for traces on npm, read-only, tenant-scoped
Claude Code flight recorder
Tamper-evident ledger, offline-verifiable
Self-host, single Compose file + Kubernetes Enterprise BYOC
License Apache 2.0 MIT (core) MIT MIT none for OSS; Enterprise self-hosted SKU is proprietary

shipped and documented  ·  supported with a real limit  ·  not documented on the vendor's own page as of 2026-09-07

† Benchmark: 2 ms p50 / 4 ms p95 / 5 ms p99 added latency on LiteLLM's own AIGatewayBench harness (4 vCPU, two runs, capture on) — full method and raw CSVs.

Not yet: SOC 2 report, configurable data residency, HIPAA/BAA, documented SAML SSO, contractual SLA, documented Slack/PagerDuty alert integrations.

Full matrix with sources → Benchmark method →

The audit ledger

Every run. On the record. Provably.

SHA-256 hash-chained, Ed25519-signed — verify any trace offline.

Full security posture →

tlane — verify
# verify any trace offline — no trust in us required
$ tlane verify <trace-id>
 signature valid (Ed25519, Merkle root)
 hash chain unbroken (SHA-256)
 tamper-evident: yes  

eu ai act

Article 12 export pack — full audit trail for high-risk AI systems. Annex III enforcement provisionally deferred to Dec 2, 2027 (Digital Omnibus political agreement, May 7, 2026; pending formal adoption — details); the ledger is available today.

india dpdp

Phase II enforcement Nov 13, 2026. Tracelane enforces no per-tenant storage region today; region pinning is on the roadmap. The audit ledger has no expiry clock — it is kept for the life of the ledger.

hipaa

No signed BAA today, so do not route PHI through the hosted gateway. Prompt and response text is not stored unless we enable capture for your workspace. Inline redaction (Team and above) masks structured identifiers only — card numbers, SSNs, emails, E.164 phone numbers and secret-shaped strings — never names, addresses or free-text clinical detail. Provider keys are envelope-encrypted.

Migration

Moving off a tool that changed hands or changed its terms?

One command rewrites your config. Nothing leaves your infrastructure.

tlane — migrate helicone
# rewrites HELICONE_* env vars, base URLs and custom
# headers across your source tree — printed as a diff,
# dry-run by default
$ tlane migrate helicone
tlane — import-litellm
# reads litellm_config.yaml, emits the
# equivalent Tracelane config
$ tlane import-litellm

Pricing

Pricing.

Six meters, priced the same way on every paid tier. Self-host free forever.

OSS self-host

$0

Apache 2.0 · forever

  • Full Rust gateway · 191 providers — 6 native adapters plus any OpenAI-compatible endpoint
  • ClickHouse + R2 storage
  • Tamper-evident audit ledger
  • Community Discord support

Free

$0

3 days indexed · 0.25 GB hot · 1 seat

  • 1 GB ingest / mo
  • 60 RPM
  • No credit card · ages out after 14 idle days, never a 429

Builder

$29/mo

30 days indexed · 5 GB hot · unlimited seats

  • 20 GB ingest / mo
  • 600 RPM

recommended

Team

$229/mo

90 days indexed · 75 GB hot · unlimited seats

  • 100 GB ingest / mo
  • 6,000 RPM
  • WorkOS SSO

Business

$799/mo

180 days indexed · 300 GB hot · unlimited seats

  • 200 GB ingest / mo
  • 60,000 RPM
  • WorkOS SSO

Enterprise

from $2,499/mo

1 year+ indexed · custom hot · unlimited seats

  • custom ingest / mo
  • Uncapped RPM
  • WorkOS SSO
  • 7-year ledger retention (included)

How the six meters work

Ingest

$0.20 / GB accepted

Every span you send, gateway or OTLP — one meter.

Hot indexed window

from $16.00 / resident GB-month

Resident GB is the logical size of what you sent, averaged daily across the month — not peak, not compressed bytes on disk. The hot window is priced as query capacity, not disk.

Series

$0.008 / series-month

Distinct metric series you emit, counted once per month.

Query compute

$0.15 / scan-unit

1 scan-unit = 1 GB scanned to answer your queries.

Cold archive

$0.08 / GB-month

Everything past your indexed window, kept queryable for two years. Metered, not unlimited: every paid plan includes a cold allowance of 24× its monthly ingest (480 GB-month on Builder, 2.4 TB-month on Team, 4.8 TB-month on Business); the rate above applies beyond it.

Evaluations

$0.005 / judge run

Each automated judge run scored against your traces.

on every tier

Tamper-evident agent ledger

The hash chain, Ed25519-signed Merkle root, Rekor anchoring and self-verification are included on every tier — free and paid alike. Seven-year retention on Enterprise. Tracelane enforces no per-tenant storage region today, signs no BAA, and makes no regulatory-conformance claim — the assessment remains yours. EU AI Act Annex III enforcement provisionally deferred to Dec 2, 2027 (Digital Omnibus political agreement, May 7, 2026; pending formal adoption — details).

"Resident GB" is the logical size of what you sent, averaged daily across the month — not peak, not compressed bytes on disk. No rollover of unused allowance. Never metered: per host or peak host, per seat on any paid tier, a separate egress line, retention as a multiplier, or OTel attributes reclassified as premium metrics. Free ages out after 14 idle days — it never returns a 429, and ingest is never blocked by billing state on any tier.

Security posture

Built for procurement reviews.

release integrity

Sigstore Cosign keyless signing with attested build provenance. Trusted Publishing OIDC — no long-lived registry tokens.

sbom

CycloneDX SBOM published with every release, verifiable from GitHub release artifacts.

vuln scanning

gitleaks, cargo-deny and a critical-level dependency audit on every push; OSV-Scanner and Grype nightly and a Syft SBOM weekly on the public repository.

pre-flight policy

Policy checks run inline at the gateway before the call leaves. Today: schema-violation + tool-definition-drift, fail-open. More on a published roadmap.

transport

TLS 1.3 minimum. mTLS via SPIFFE/SPIRE for ingest. SSRF defense — RFC1918 + link-local denied.

standards

OTel GenAI semconv — tracking the pre-execution judgment / negative-proof proposal (#3244).

FAQ

Frequently asked.

What is Tracelane?

An open-source (Apache 2.0) flight recorder for AI agents: a universal Rust LLM gateway, full-fidelity OpenTelemetry observability, and a tamper-evident audit ledger. Each event is SHA-256 hash-chained and the Merkle root is Ed25519-signed every 100 events, so any trace can be verified offline.

How is it different from Helicone, LangSmith or Langfuse?

Tracelane unifies the gateway, observability, pre-flight policy and a tamper-evident audit ledger in one Rust binary. Helicone, Langfuse and LangSmith are observability-first — you bring your own gateway and guardrails. Licensing: Tracelane is Apache 2.0 core with a published self-host path; Helicone's gateway is GPL-3.0, Arize Phoenix is ELv2 (no SaaS resale), LangSmith is closed-source. The audit ledger is hash-chained and Ed25519-signed — we are not aware of another product in this category that offers cryptographic audit-log integrity at the trace level.

What about LangSmith Engine?

LangSmith Engine (May 13, 2026) clusters production failures after they happen and drafts pull requests to fix them; Tracelane's pre-flight policy checks flag failure modes at the gateway. Engine is locked to the LangSmith trace surface and LangChain-managed inference keys per LangChain's terms of service; Tracelane is neutral OTel ingest with 0% BYOK markup across 191 providers. Different layer of the agent reliability stack.

Is Tracelane really open source?

Yes. The core (Rust gateway, ingest workers, observability backend, pre-flight layer, CLI, SDKs, audit verifiers) is Apache 2.0. The /spec directory is CC0. Paid features are gated by the workspace_entitlements table at the application layer, not by a license-string compare.

Which providers are supported?

191 providers today — 6 native adapters (Anthropic, Google Gemini, Google Vertex, AWS Bedrock, Azure OpenAI, Cohere) plus OpenAI and any OpenAI-compatible endpoint, including Together, Fireworks, Groq, OpenRouter, DeepSeek, xAI, and Mistral. One canonical wire format across all of them.

Does Tracelane support EU AI Act Article 12?

The Tamper-Evident Agent Ledger (SHA-256 hash chain, Ed25519-signed Merkle root, Rekor anchoring, self-verification) is included on every tier, free and paid — no add-on to buy. The Article 12 evidence export and seven-year ledger retention are included on Enterprise. Tracelane enforces no per-tenant storage region, signs no BAA, and makes no regulatory-conformance claim. EU AI Act Annex III obligations were originally scheduled for Aug 2, 2026; the Digital Omnibus political agreement (May 7, 2026) provisionally deferred enforcement to Dec 2, 2027, pending formal adoption (details). The ledger ships now so regulated buyers can build audit history before binding enforcement.

What's the latency overhead?

Measured, not targeted: 2ms p50, 4ms p95, 5ms p99 of added latency (excluding provider time), taken with LiteLLM's open AIGatewayBench harness on a 4 vCPU host with capture on, two runs, alongside Bifrost, Portkey and LiteLLM — see the full breakdown above or the benchmark page for method, caveats and the raw CSVs. The pre-flight policy layer is still an engineering target (p99 <50ms on commodity CPU).

Can I self-host?

Yes. Docker Compose for development, Helm chart for production. The full ClickHouse + R2 stack runs on a single small Hetzner node. See the self-hosting guide.

Is Tracelane available now?

Yes — V1 is live. Start free at app.tracelane.dev, or self-host the Apache 2.0 core.

Stay close

Get product updates.

One email when something ships. Unsubscribe in one click.