langfuse.com
@langfuse
Followers
1,542
Following
499
Media
72
Statuses
387
Open source LLM engineering platform. Traces, evals, prompt mgmt and metrics to debug and improve your LLM app. We're hiring:
github.com/langfuse/langfuse
Joined July 2023
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
seokjin
• 512297 Tweets
Nebraska
• 150506 Tweets
ホームラン
• 114733 Tweets
Noroña
• 60628 Tweets
ENCOURAGE CHARLOTTE21
• 47864 Tweets
大雨特別警報
• 37759 Tweets
ロッキン
• 33474 Tweets
From Russia with Lev
• 32544 Tweets
#キントレ
• 29426 Tweets
能登半島
• 26222 Tweets
SO KISS ME LISA
• 26095 Tweets
DAOUOFFROAD in MACAU
• 20738 Tweets
レイジョ
• 16985 Tweets
仮設住宅
• 10562 Tweets
Rishabh Pant
• 10083 Tweets
Pinned Tweet
⭐️ 5-0-0-0 🤩
We've just crossed 5.000 stars on GitHub. Thank you!
Very proud of our recent progress:
• GPT-4o mini support for tokenization and cost tracking
• Trace timeline view in the Langfuse UI
• Integrations with
@dify_ai
,
@Haystack_AI
,
@mirascopeai
and
@milvusio
2
2
20
We’ve raised a $4M seed round led by
@lightspeedvp
and
@lafamigliaVC
with participation from
@ycombinator
.
11
17
123
😻 We're back on
@ProductHunt
- where we originally launched
@langfuse
last year.
🆙 Check out our launch and lend us your support:
👇 Marc is introducing Langfuse 2.0 and running through all new features in the video below
6
5
58
Thanks everyone for the Golden Kitty Award 🏆
Langfuse has come a long way since we initially launched on
@ProductHunt
last summer and we are incredibly grateful for the love and support by our community!
7
4
50
🧠 Pleasant surprise to be included in
@thoughtworks
latest tech radar!
It's here!
The 30th edition of Thoughtworks Technology Radar is live!
15 years, 30 editions – thanks for being part of the journey.
Read now:
2
41
129
2
6
33
We're hiring a founding engineer to join us in building out OSS LLM Observability
1
6
23
langfuse's code is open source and now our roadmap is, too
1
0
21
🥇 Thrilled to have taken home Product of the Day on
@ProductHunt
on Friday.
😻 As ever, a day full of energy and excitement for us. It was a ton of fun to share this with our users and other makers on the platform.
🔙 We'll be back!
2
4
17
😻 we're nominated in AI infra & open source! lends us an upvote: links below
🚨 It's tiiiiiiiiiime
Golden Kitty Awards 2023 voting is now OPEN!
Who's going to purr their way to the top this year? Vote now for the products that shaped your year
6
2
24
2
1
17
We're live on
@devhunt_
this week🪢 We've been impressed by the quality of submissions of the past months + love that it's OSS. Give us a looksy! 💻
@johnrushx
2
3
17
🖨️ hot off the press: tag traces in
@langfuse
assign tags via SDK or in the UI to slice and dice your traces to your liking. assign language, tool usage, environment, review tasks, golden data etc.
changelog:
first feature by the talented
@RichardKrue
,
1
2
13
Launching: Drop-in replacement of OpenAI SDK to get started with Langfuse super-⚡️
1
2
15
Hard to believe that we've been too busy to share that
@brexHQ
put us on a billboard in Times Square in October!
Pretty surreal.
#GrowWithBrex
@conorbradd
2
1
14
Langfuse Update — August 2023
◆ Langchain integration for JS/TS
◆ Track releases and versions of an LLM app, chain or prompt
◆ Cost in USD for many models
◆ Simplified self-hosting
◆ ... many small improvements and bug fixes
Read on 👇
1
2
13
New cookbook: adding Langfuse tracing to
@vercel
AI SDK
This makes it easy to stream from the edge while using Langfuse to evaluate, debug and continuously improve the application
0
2
13
Langfuse now has native OSS tracing of all LangChain interfaces
✅ invoke/run/call/...
✅ async
✅ batch
✅ streaming
✅ Python & JS
Learn more 👇
1
2
13
We're rolling out prompt management in beta today. DM if you want to test the feature (manage prompts from langfuse UI or SDK, keep code lean, get version data, run experiments)
3
0
12
New integration with
@Haystack_AI
🌾OSS Observability for
@Haystack_AI
with
@langfuse
🔮 Build RAG applications using Haystack 2.0's clean abstractions and easily trace, evaluate and debug them in Langfuse.
Add the `langfuse-haystack` package, enable `HAYSTACK_CONTENT_TRACING_ENABLED` and all Haystack
1
5
16
1
0
12
We are very excited to launch our integration with langflow by
@logspace_ai
today. More below in the thread:
1
3
11
🪢 Happy to share that
@langfuse
is a
@ycombinator
top 100 open source & gen AI startup of 2024.
🚢 now back to work
1
2
8
Evals are fast becoming one of Langfuse's most adopted features after core observability.
When logging a lot of production usage to Langfuse, teams start layering model-based evals on top of the manual checks and reviews to scale their evaluation.
1
2
11
We are proud to help developers craft new product experiences with LLMs and see them launch on
@producthunt
.
Congrats
@touring_app
,
@gab_creates
& team!
And thanks for the shoutout alongside
@elevenlabsio
and
@GroqInc
2
3
11
LLama 3 just dropped - exciting day for open source~
0
0
11
New
@dify_ai
x Langfuse integration
Build and host your LLM applications with
@dify_ai
, monitor, evaluate and debug with
@langfuse
See docs below 👇
💥 Dify v0.6.12 has landed!
We are thrilled to announce that Dify has integrated LangSmith and Langfuse as additional LLM application observability solutions.
This integration enables you to trace and evaluate the quality and cost of your LLM applications built with Dify.
You
6
33
96
2
6
10
Did you know...?
You can use hotkeys to navigate between sessions, users and traces in Langfuse. Press `j` to navigate down and `k` to navigate up. Doing our part in accelerating 😎
-- Keep your eyes peeled, we'll be sharing more nifty features over the holidays.
0
1
10
Launch Week Day 2: LLM Playground
Introducing:🛝 LLM Playground on
@langfuse
🫳 Day 2 drop of Langfuse Launch Week: We extend the scope from observability, evaluation and prompt management to interactive exploration within the LLM playground.
Since introducing prompt management, many product managers and
3
0
37
1
0
10
Self-hosting Langfuse just got easier
Just pull the latest Docker image and all pending migrations are applied automatically ✨
See changelog for full details:
0
2
10
Thanks for featuring us in the Enterprise AI market map!
@weekendfund
,
@vedikaja_in
,
@rrhoover
We're open source, and Enterprises can self-host Langfuse (dockerized) for observability and analytics of LLM apps! Reach out for details on support plans.
🤫 PH launch coming up
2
2
9
Launch Week Day 0: OpenAI JS SDK Wrapper
Our OpenAI SDK wrapper for Python has been very popular. Today we are happy to generally release the JS port.
Tracking of:
✅ prompt/completion
✅ api errors
✅ tokens / USD cost
✅ latency and time to first token
✅ OSS
6
1
9
@posthog
is the company that inspires us most in our day-to-day. It was delightful to sit down with
@JoeMart71595198
and share our story. Many pivot and
@ycombinator
learnings to be gleaned …
2
1
8
😲 We've been impressed with
@mirascopeai
since we first met them a week ago!
🫳 A pythonic way to build LLM apps just got more even better: a native, decorator-based integration with Langfuse just dropped.
👇 Video Walkthrough below
1
2
9
Launch Week Day 3: observe() Decorator for Python in
@langfuse
🪴 observe() Decorator for Python in
@langfuse
🫳 Day 3 drop of Langfuse Launch Week: The Decorator is the new recommended way to use Langfuse in Python. All inputs, outputs and timestamps are captured automatically. Using decorators makes nesting and logging complex traces a
0
1
18
0
0
9
Want to learn more about the technical details of the Langfuse Decorator? We published a short blog post covering some of the more interesting design decisions.
0
1
9
⭐️ 3-0-0-0 🤩
We've just crossed 3.000 stars on Github. Thank you to each and every one of you!
3
1
9
The
@hamilton_os
python framework is looking for users who'd want to explore a
@langfuse
integration. They're based on DAGs, runnable in Python which allow for complex, tree-like LLM executions. Reach out to
@stefkrawczyk
2
4
9
Keenly attentive users will have noticed our Sessions feature dropping in beta late last week. Expect an announcement this week.
Until then for the curious:
0
0
8
Did you know? Shareable Links
You can share traces publicly with people that are not part of your project. Just toggle the 'Public' Button at the top right and copy the URL by clicking.
0
1
8
✨ Pretty-mode for common LLM I/O
Starting with OpenAI Chat Messages
1
2
8
🚢
⛓️🦜
Langfuse x LangChain JS/TS integration is hot off the press ☕️
1
1
8
😽 We're topping of Launch week with a
@producthunt
launch tomorrow (Friday)
🫳 + there will be our biggest feature drop of the week.
1
3
8
@MarcKlingen
will reply to you on github discussions 🤓
Our github discussions are positively bopping -- we started using them two weeks ago. Already finding them more useful than discord: they create an artifact that can be easily found and tagged/moderated. Great for feature requests and support.
2
1
10
1
2
4
New integration: 🦙
@llama_index
x 🪢 Langfuse
Launching the
@langfuse
🪢 <->
@llama_index
🦙 integration
@langfuse
is open source observability for RAG: understand retrieved context, latencies, costs & quality of responses + identify hallucinations
Docs:
integrate in 1-line of code --> Details 👇
2
9
53
1
1
7
Keeping up with as the team 🚢
Changelog: Support for ChatML (role/message) prompts in Langfuse prompt management including fully typed SDKs and client-side caching
0
1
6
0
2
7
Langfuse SDK now caches prompts client-side for increased performance with custom TTL 🚀
1
0
7
Today, Langfuse is a thriving open source project supporting developers of LLM applications throughout the development lifecycle of their applications.
1
0
7
@brexHQ
is our partner since day 0. Thank you for showcasing us in SF last week during
@TechCrunch
disrupt
@conorbradd
🕵️♂️ Langfuse (YC W23) is Sherlock Holmes for your LLM app: We help you investigate what your LLM app does in production.
1
1
6
Launch Week Day 1:
@posthog
integration
0
0
6
Launch Week Day 5: Model-based Evaluation
Introducing: 💡 Model-based Evaluation in
@langfuse
🫳 Final drop of Langfuse Launch Week: Model-based Evaluations within Langfuse allows for continuous evaluation across various dimensions like hallucinations, toxicity, relevance, and more.
With Langfuse-managed templates and
1
3
15
0
0
6
🫳 launch week 1 is in the books
🆕 langfuse 2.0 is live
👇 summary below
2
1
6
🎙️ Learn more about tracing and observability for LLMs.
📺 Webinar tomorrow at 8am PT / 5pm CET on
"Traceability and Observability in Multi-Step LLM Systems"
@marcklingen
will speak with
@PhilipTannor
@llmopsSpace
- grab a popcorn and register below👇
2
2
4
Changelog: Support for ChatML (role/message) prompts in Langfuse prompt management including fully typed SDKs and client-side caching
0
1
6
We're nominated for a
@ProductHunt
Golden Kitty Award
Vote Langfuse in the open source category:
1
0
6
UI improvement: "Pretty ✨" mode for all ChatML traces that include additional params like functions/tools
0
1
6
Thank you
@FouquetSanne
for shouting out Langfuse on the
@alltrendsai
newsletter. We love that
@joinmelonio
runs on Langfuse and everybody should be checking it out to power their second (LLM-based) brain 🧠
3
2
6
🚨 Event Alert - Join us this Thursday when
@marcklingen
speaks with
@PhilipTannor
about implementing traceability and observability in multi-step LLM systems.
⏰ Thursday, 8am PT, 5pm CET
👇 Link to register below
1
5
7
accelerating the accelerators
Did you know...?
You can use hotkeys to navigate between sessions, users and traces in Langfuse. Press `j` to navigate down and `k` to navigate up. Doing our part in accelerating 😎
-- Keep your eyes peeled, we'll be sharing more nifty features over the holidays.
0
1
10
0
0
6
We love to turn around little improvements within hours. All the more motivating to receive feedback like this.
1
0
6
Did you know? Exporting from Langfuse
You can search, filter and export generations as .csv, JSON and JSONL (for finetuning).
The exports include information such as trace ID, name, trace name, latencies, cost and model and allow you to run custom and flexible analysis on your
0
0
5
@vanstriendaniel
💥💥💥
The
@langfuse
@OpenAI
integration is super convenient and can be used with any other model that uses the same API (e.g.
@MistralAI
and
@perplexity_ai
or many more via
@LiteLLM
)
0
1
5
(1/3) Many teams building complex LLM applications iterate a lot on it. With Langflow's intuitive UI, you can effortlessly build even complex LLM applications in no-code to move fast.
⛓️
1
1
5
New integration:
@FlowiseAI
🤝 Langfuse
New release update!
Flowise v1.3.5 with:
💬 Fine-tuned ChatGPT
🌴 Google PaLM Chat & Embedding
🔍 Langfuse support
🛑 Rate Limiting
📚 Vectara Files Upload
🌟 Fixes and upgrade of existing nodes
6
4
55
0
0
5
Read the full story on our blog
1
0
5
👨🎨 Now with a slick UI courtesy of
@RichardKrue
👾 Now rolling out prompt management beta to all Langfuse users.
🕹️ See below to get started
1
0
4
0
0
5
We're alpha testing a
@LangChainAI
integration for our observability platform
- get in touch if you'd like to test + give feedback.
0
3
5
(3/3) In Langfuse, all the Traces recorded in the Langflow app become visible in real time. From here, you can debug or access analytics.More on the integration:
0
1
5
Day 1 support for OpenAI `gpt-4o` including new tokenizer (`o200k_base`) for accurate token counts.
0
2
5
strong start this monday
stargazer slack channel became a bit busy – here we go again ❤️
0
3
13
0
0
5
Launch Week Day 4: Datasets v2
Introducing: 💾 Datasets 2.0 - Manage fine-tuning, testing, and evaluation datasets in
@langfuse
🫳 Day 4 drop of Langfuse Launch Week: Datasets get a major update. We left no stone unturned - new tables, a new editor, more API endpoints for full programmatic access, metadata
1
3
14
0
0
5
tinyllm 🤝 Langfuse
🪢⚡️🪢⚡️
🚨Sharing tinyllm, a Python framework for LLM/Agent based applications.
⚡Build, monitor and evaluate LLM apps at scale
⚡Trace and debug LLM apps with
@langfuse
Currently using tinyllm as infra for my AI powered Finance app !
0
2
2
0
1
4
Changelog: OpenAI SDK integration (Python) now captures used prompts from Langfuse prompt management
Thanks to
@fancyweb_
for the contribution 🙏
1
0
4
Who doesn't ❤️ cookbooks
Introducing new section in
@langfuse
docs
Featuring all end-to-end cookbooks, sometimes it is just easier to understand an example than to read the docs
Personal highlight: directly open notebooks on google colab or download from github
0
1
7
0
2
4
Bringing Langfuse to the next audience: Great to see an integration with
@Langflowai
merged today
0
1
4
Demo alert
Interactive langfuse demo is now live. Give it a spin here:
0
0
4
and we're live:
0
0
4
Thrilled to be launching langfuse & announcing our backing by
@ycombinator
Congrats on the launch,
@rawert
,
@marcklingen
&
@maxdeichmann
!
@Langfuse
(YC W23) is building open-source product analytics for LLM apps, helping companies to track and analyze the quality, cost, and latency of LLMs, chains & agents in granular detail.
2
2
35
0
0
4
(2/3) After building the app, you can start using it immediately from the UI and see the results.
1
1
4
👾 Now rolling out prompt management beta to all Langfuse users.
🕹️ See below to get started
Last drop in 2023, manage prompts in
@langfuse
🎇
Collaboratively iterate on prompts, promote them to production from UI, and track metrics by prompt version.
Learn more:
Performance, analytics, and ui improvements planned for next releases.
4
2
23
1
0
4
Check out our live demo to give Langfuse a spin
0
0
4
By popular demand: Sign into Langfuse with Azure AD🛡️
Often requested by our enterprise self-hosting users. Great to get this out of the door on Christmas Eve
@MarcKlingen
@azuread
@Azure
0
0
4
We've been busy in July. Looking forward to continue to 🚢 in August
🚢 Team is on a roll, excited to share what we've shipped to
@langfuse
(open-source) in July:
◆ New Trace UI
◆ Human-in-the-lop eval
◆ User-level reports
◆ Langchain integration
◆ Improved SDKs
◆ Automated token counts
Details/screenshots 🧵
2
2
7
0
0
3
New Claude models supported in Langfuse's cost and token usage tracking (+ flexibility to add any model and pricing)
0
1
3