GILD verdict
Choose Langfuse to run the whole platform yourself or score most traces, and Braintrust when a managed eval product matters more.
- The workflow
- an AI team that scores production traces and gates releases on evals, deciding where traces live and how the bill grows
- The decisive evidence
- ReportedCREATE SOMETHING replaced Braintrust with Langfuse across its evals, telemetry and CI workflows on Jul 6, 2026 and reran nine eval suites against Langfuse. No named team in GILD's research moved the other way.1
- DocumentedAt 200,000 traces a month with 10 observations and one score each, GILD's calculation from published rates comes to US$369 on Langfuse Pro and US$474 on Braintrust Pro, where 150,000 extra scores cost US$225.2412
- DocumentedLangfuse self-hosts the whole platform free under the MIT license. Braintrust self-hosts only its data plane, on Enterprise, while it runs the UI and control plane.314
- GILD analysisWhy teams pick one over the other is unresolved. CREATE SOMETHING's pull request states no reason for its move, and no named team has published a move from Langfuse to Braintrust.
- The tradeoff
- Braintrust fits while your scores per trace stay few. Langfuse bills every observation, so deep agent traces with few scores can cost more there than on Braintrust's per gigabyte rate. Braintrust gives you a managed eval workflow but no full self-hosting below Enterprise.341214
- What would change the call
- The call moves when a team publishes why it chose Braintrust over Langfuse on its own traffic, or when either vendor changes the per observation or per score rate behind the bill.
Langfuse is the one you can run entirely yourself: it self-hosts free under the MIT license, and Braintrust self-hosts only its data plane, on Enterprise. CREATE SOMETHING replaced Braintrust with Langfuse across its evals, telemetry and CI workflows on July 6, 2026. At 200,000 traces a month with one score each, GILD's calculation from published rates comes to US$369 on Langfuse Pro and US$474 on Braintrust Pro; payload size and score count move that gap.
What teams report
Switched 1
CREATE SOMETHINGOn Langfuse
A pull request merged on Jul 6, 2026 replaced Braintrust across the monorepo's evals, telemetry, packages, scripts and CI workflows with Langfuse, then reran nine eval suites against Langfuse. It states no reason for the move.
“…replace active Braintrust eval, telemetry, package, script, env, and workflow surfaces with Langfuse equivalents…”
createsomethingtoday, Repository owner, CREATE SOMETHINGGitHub (createsomethingtoday/create-something-monorepo), , post
Choosing between Langfuse and Braintrust
Choose Langfuse if
- You must self-host and want the whole platform, control plane included, free under the MIT license.
- Your traces are scored often: at one score per trace and 200,000 traces a month, Langfuse Pro comes to US$369 against US$474 on Braintrust Pro in GILD's calculation from Oct 28, 2026 rates.
- You want enterprise SSO without an enterprise contract: it is included in self-hosted OSS, and in Cloud through the US$300 Teams add-on on Pro.
Choose Braintrust if
- Evaluation is the main job and you want a managed product built around experiments, scorers, and playgrounds.
- Your traces are deep but light and your scores few, so billing by gigabyte processed and score beats paying for every observation.
- You want your data plane in your own cloud while Braintrust runs the UI and control plane, on Enterprise through BYOC or self-hosting.
Side by side
| Dimension | Langfuse | Braintrust |
|---|---|---|
| What teams report | CREATE SOMETHING replaced Braintrust's eval, telemetry, package, script and workflow surfaces with Langfuse and reran nine eval suites against it, merged Jul 6, 2026. The pull request gives no reason.1 | No named team in GILD's research moved from Langfuse to Braintrust as of Oct 28, 2026. |
| PlansList prices in US dollars. Braintrust includes unlimited users on every plan; Langfuse from Core up. | Hobby free with 50,000 units a month and 2 users; Core US$29 a month; Pro US$199 a month; Enterprise US$2,499 a month; self-hosting free under the MIT license23 | Starter US$0; Pro US$249 a month; Enterprise custom, invoiced annually12 |
| Billing unitDeep traces with small payloads cost more on Langfuse; large payloads with few scores cost more on Braintrust. Price a week of your own traffic on both. | A unit: each trace, each observation inside it, and each score, including scores from Langfuse's own evaluators4 | Gigabytes of processed data, measured at ingestion, plus scores; built-in model use draws on monthly credits12 |
| Included volume and usage price | 100,000 units included on Core and above, then US$8 per 100,000 to 1 million, US$7 to 10 million, US$6.50 to 50 million, and US$6 above2 | Starter: 1 GB and 10,000 scores a month, then US$4 per GB and US$2.50 per 1,000 scores. Pro: 5 GB and 50,000 scores, then US$3 per GB and US$1.50 per 1,000 scores12 |
| 200,000 traces a month, 10 usersGILD's calculation from published rates, Oct 28, 2026. Inputs: 200,000 traces a month, 10 observations and 1 code score per trace, 20 KB per trace, 10 users, 30 days retained, no judge calls. | US$369 a month on Cloud Pro: US$199, plus US$72 for the first 900,000 units past the 100,000 included at US$8 per 100,000, plus US$98 for the next 1.4 million at US$7. Each trace is 12 units, 2.4 million in all.24 | US$474 a month on Pro: US$249, plus US$225 for 150,000 scores past the 50,000 included, at US$1.50 per 1,000. The 4 GB of data stays inside the 5 GB included.1213 |
| Retention | Data access for 30 days on Hobby, 90 days on Core, and 3 years on Pro and Enterprise. Project retention policies from Pro up, minimum 3 days; self-hosted data is kept until you set a policy5 | 14 days on Starter; 30 days on Pro, extendable to 180 days at US$0.50 per GB a month; custom on Enterprise; experiments up to 365 days1216 |
| Self-hosting | Free under the MIT license with all core features and unlimited usage; an Enterprise key adds project RBAC, retention policies, audit logs, and server-side masking36 | Enterprise only. The data plane runs in your AWS, GCP, or Azure account through Braintrust's Terraform modules, and Braintrust hosts the UI and control plane. With BYOC, Braintrust operates it in your cloud14 |
| Evaluation in CI | Experiments via SDK or UI, and the langfuse/experiment-action GitHub Action, which fails a pull request when a score crosses its threshold9 | Evaluations in code, the UI, or CI/CD, recorded as experiments to compare across runs17 |
| SSO and access control | Cloud: enterprise SSO and project RBAC through the US$300 a month Teams add-on on Pro, included on Enterprise. Self-hosted OSS includes enterprise SSO27 | Google sign-in on Starter and Pro; SAML or OIDC SSO and custom permission groups on Enterprise1215 |
| Compliance | SOC 2 Type II and ISO 27001 reports from Pro, a BAA in the HIPAA region from Pro, and a GDPR DPA on every plan8 | SOC 2 Type II attestation and a BAA on Enterprise; click-through DPA on Pro15 |
| License | MIT for the core, with Enterprise features under a separate license in the ee directories7 | Proprietary platform; the autoevals scorer library is open source |
What practitioners report, side by side
Each product's figures come from its own practitioners, for the group you pick.
Overall signal
Langfuse reads +46 and Braintrust +50, each from its own practitioners. Langfuse: 12 positive, 0 negative of 26. Braintrust: 20 positive, 3 negative of 34.
The score counts the authors from teams and leaders that take a side, one vote each with their latest view, and is how far positive votes lead negative ones, per 100 votes, so Braintrust's +50 rests on votes split 19 positive, 10 mixed and 3 negative. Each of Langfuse's authors wrote one source here, so each source is one vote.
A lead of a fifth of the votes reads positive and a lead of half strong positive, with the same lines below zero. The dots show each source, positive, mixed, negative. Run a pilot on one production app to measure which failures it surfaces and how long tracing them takes.
Footprint on the AI Application Platform stack
They differ on two of 9 layers. On Model gateway, Langfuse touches it and Braintrust supports it. On Tool use, Langfuse touches it and Braintrust supports it. Coverage is each product's placement on this stack, as its product page shows it. Core is a job the product owns; Supports and Touches are jobs it helps with. Coverage says nothing about how well a product does them.
| Layer | Langfuse | Braintrust |
|---|---|---|
| Model | ||
| Model gateway | Touches | Supports |
| Build | ||
| Tool use | Touches | Supports |
The 7 layers where they match
| Layer | Langfuse | Braintrust |
|---|---|---|
| Model | ||
| Model APIs | Touches | Touches |
| Build | ||
| Orchestration | Touches | Touches |
| Data | ||
| Retrieval | Touches | Touches |
| Evaluate | ||
| Evaluation | Core | Core |
| Operate | ||
| Observability | Core | Core |
| Cloud deployment | Touches | Touches |
| Guardrails | Touches | Touches |
What practitioners praise
Among teams and leaders, Datasets, Experiments & Scorers and Instrumentation & SDK Coverage reach 5 sources from 3 authors only for Braintrust, at 20 of 21 and 3 of 5. Langfuse needs a pilot on the same production traffic before you choose on either. Both carry Trace-Level Debugging. It is praised in all 5 Langfuse sources that discuss it, against 7 of 8 for Braintrust.
| Praised for | Langfuse | Braintrust |
|---|---|---|
| Trace-Level Debugging | 5 of 5 | 7 of 8 |
| Datasets, Experiments & Scorers | Not a theme in its sources | 20 of 21 |
| Instrumentation & SDK Coverage | Not a theme in its sources | 3 of 5 |
Each count is the sources that praise a theme, or raise it as a concern, out of those that discuss it. A row pairs the two products when both carry a theme under the same or an equivalent name.
A count ending in "so far" has under 5 sources from 3 authors among teams and leaders. "Not a theme in its sources" means GILD found no theme of that name in that product's sources.
What practitioners raise as concerns
Among teams and leaders, SDK & Ingestion Reliability and Build It Yourself reach 5 sources from 3 authors only for Braintrust, at 6 of 6 and 2 of 6. Langfuse needs a pilot on the same production traffic before you choose on either. Upgrades & Migrations reaches 5 sources from 3 authors only for Langfuse, at 5 of 6. Braintrust needs a pilot on the same production traffic before you choose on it.
| Raised as a concern | Langfuse | Braintrust |
|---|---|---|
| Upgrades & Migrations | 5 of 6 | Not a theme in its sources |
| SDK & Ingestion Reliability | Not a theme in its sources | 6 of 6 |
| Build It Yourself | Not a theme in its sources | 2 of 6 |
18 sources
- createsomethingtoday/create-something-monorepo pull request #783: retire Braintrust for Langfuse (Jul 6, 2026)
- Langfuse pricing
- Langfuse self-hosted pricing
- Langfuse docs: Billable units
- Langfuse docs: Data retention
- Langfuse docs: Self-host Langfuse
- Langfuse docs: Enterprise license key (self-hosted)
- Langfuse security and compliance
- Langfuse docs: Experiments in CI/CD
- Langfuse docs: Evaluation overview
- Langfuse docs: OpenTelemetry
- Braintrust pricing
- Braintrust docs: Plans and limits
- Braintrust docs: Self-hosting Braintrust
- Braintrust docs: Security
- Braintrust docs: Data retention
- Braintrust docs: Run evaluations
- Braintrust docs: Send OpenTelemetry traces and logs