Laminar flow-1 Pricing: Signals Costs and Cloud Plans
Laminar launches flow-1 trace analysis. Compare its published rates, the vendor's 23× cost claim, and what agent teams should test before switching.
By AI Pricing Guru Editorial Team
The AI Pricing Guru editorial team verifies prices against official provider pages and maintains the datasets, calculators and guides. See our methodology and contact us to report a correction.
Report summary — 2026-10-19
- Laminar's October 19 Launch HN spotlights flow-1, a specialized model that investigates agent traces through Signals, with published token rates.
- Laminar reports 23× lower analysis cost than GPT-6 Sol on its own benchmark; that is not an independently verified saving or a cut to your agent's API rates.
- Pilot it on labeled production failures: flow-1's higher precision came with lower recall than Sol in the reported test. Open-source tracing does not make managed analysis free.
| Plan | Monthly base | Data / overage | Signals |
|---|---|---|---|
| Free | $0/mo | 1 GB No overage | Unresolved: pricing says no; guide offers trial credit |
| Starter | $30/mo | 3 GB $2/GB extra | Usage billed + $5 one-time credit |
| Pro | $150/mo | 10 GB $1.5/GB extra | Usage billed + $5 one-time credit |
| Enterprise | Custom | Custom Custom | Custom |
Prices, limits and billing verified ; Free Signals access unresolved · USD · Official source
flow-1 Signals analysis rates
Per million analysis tokens: $0.05 input · $0.01 cached input · $0.3 output. These are the Signals agent’s analysis tokens, not your monitored agent’s tokens. Paid cloud base and data fees are separate.
Data access: Free 7 days · Starter 30 days · Pro 6 months. Free: one project and one seat; paid: unlimited projects and seats.
Try Laminar · Platform pricing API · flow-1 availability API
Laminar's reported cost per 1,000 analyzed traces
Vendor benchmark, not an independent test or guaranteed bill. Costs include the full Signals investigation; the cost subset uses traces below 100K total LLM tokens.
Scaled from Laminar's rounded per-trace figures. The reported 23× claim uses traces per dollar; rounding means those figures are not exact reciprocals. Benchmark and methodology →
Historical story text, scoped to the publication date above:
Laminar, a YC S24 company building open-source agent observability since 2024, presented its recently released flow-1 trace-analysis model in an October 19 Launch HN post. The actionable pricing news is its published Signals analysis rate card, not a new discount on OpenAI or Anthropic inference.
The official flow-1 announcement is dated October 2. October 19 is the Launch HN date; no previous flow-1 rate or October 19 price cut is established.
What launched—and what is open source
Signals investigates an agent’s trace using a prompt and a user-defined JSON schema. It treats spans like files, searches for evidence, reads relevant steps and returns structured findings. Similar findings are clustered so teams can identify recurring failures and turn them into evaluation datasets.
It catches mistakes ordinary uptime monitoring misses: successful tool calls can still produce unsupported claims or false completion reports.
The platform repository is Apache-2.0 licensed and self-hostable. That does not establish public availability of flow-1 weights, or inclusion of hosted Signals analysis in self-hosted deployments.
What the pricing means for agent teams
Potential winners: teams collecting more traces than humans can review. Signals bills investigator tokens, including retrieval steps—not simply the original agent’s tokens. Cloud plans, storage and original inference remain separate.
The trade-off: in Laminar’s 523-trace benchmark, flow-1 slightly exceeds Sol’s detection F1, but Sol has substantially higher recall. Prioritizing fewer missed failures may lead to a different choice than prioritizing fewer false alarms.
The 23× claim is not a universal token-price ratio or full-platform saving. Laminar also shows flow-1 missing a contract omission that both GPT-6 models catch.
Neither OpenAI pricing nor Anthropic pricing changes through this launch. Our API cost guide covers all-in budgeting.
Cloud plans and Labs availability
The plan table above renders from maintained platform pricing data: subscriptions, included data, overage, one-time credit and data-access duration. Free Signals access remains unresolved: the pricing card excludes it, while the guide describes a one-time workspace trial credit. Do not assume either source overrides the other; a one-time credit is not a recurring monthly allowance. Storage is not proportional to trace tokens. The calculator uses billed analysis tokens, with cached and uncached input as non-overlapping buckets.
Independent Labs replay is blocked. No authorized Signals project, pinned trace-analysis harness, matched test traces, raw investigations and complete billing ledger are available. The Labs coverage decision and availability API record what is needed. No paid analysis was run for this release.
What to do now
- Create a labeled pilot. Include known failures, clean runs and high-consequence edge cases. Measure precision and recall separately; manually check the linked evidence.
- Start with one Signal. A concrete rule such as “the final report contradicts tool output” is easier to validate than an unrestricted quality judgment. Use sampling and spend limits before expanding coverage.
- Measure the actual analysis bill. Record uncached input, cached input, output, investigations per trace and human review time. Use the inline calculator for actual analysis-token scenarios and our token calculator for the original agent’s separate API spend.
- Check deployment boundaries. Confirm what your cloud plan includes, retention and storage charges, and whether required analysis features work in your intended self-hosted setup.
See the full Laminar pricing page. For hosted open models powering the underlying agent, compare Novita’s inference options; inference hosting is separate from trace analysis.
Affiliate disclosure: We may earn a commission from the sponsored Novita link at no extra cost to you.
Sources and verification
Primary sources checked October 19, 2026: Launch HN, flow-1 announcement and benchmark, official pricing, Signals documentation, and platform source code. We verified the published rate card against two Laminar pages; we have not independently reproduced its benchmark.