English
Pydantic
Report

Pydantic

@pydantic

The Pydantic Stack: Logfire, Monty, Validation, AI, Evals, and AI Gateway

177 Following21.1K Followers2.9K Posts
  1. Pydantic@pydantic
    One Jev request, three scores: a policy label, a rubric level, and the probability the reply asks for a password, all recorded in Logfire. Modeled cost of 1M evaluated replies: $62, vs $4,542 with Braintrust before data charges. Blogpost
    66525.1K
  2. Pydantic@pydantic
    Pydantic
    RT @dlewis: We had a 30% reduction in tokens with our agents after we implemented a code mode with Monty. Also allowed us to scale to handle some tools that produce some very long output, so now the agent can just write filtering code in code mode to deal with it. Monty is an awesome solution!
    Pydantic @pydantic
    If your agent updates 200 records with 200 tool calls, the model writes out every call and reads back every result. With code mode, it writes one Python loop. Across 8 models and 5,276 runs, long tasks ran 51% faster and succeeded 16 points more often. Check out the report:
    03087
  3. Pydantic@pydantic
    If your agent updates 200 records with 200 tool calls, the model writes out every call and reads back every result. With code mode, it writes one Python loop. Across 8 models and 5,276 runs, long tasks ran 51% faster and succeeded 16 points more often. Check out the report:
    Download
    66605.8K
  5. Pydantic@pydantic
    Coding agents do real work, but leave little record beyond commits and PRs. Tomorrow, we'll how to trace yours in Pydantic Logfire and use those traces to compare harnesses on cost and quality. Register now:
    Download
    12161.1K
  6. Pydantic@pydantic
    Pydantic
    RT @marcelotryle: Uvicorn now speaks HTTP/2 ⚡ - Opt in with uvicorn main:app --http zttp --http2 - Negotiated over TLS with ALPN, or cleartext with prior knowledge - Response trailers and 103 Early Hints
    040105
  8. Pydantic@pydantic
    The @AgenticAIFdn gave @Marcelotryle its Outstanding Technical Contribution Award for maintaining the MCP Python SDK. He wrote a short thank you to the people who made it possible.
    Download
    34281.8K
  9. Pydantic@pydantic
    Pydantic
    RT @juanse_VA: I’ll be in Amsterdam this week for the first FastAPI Conference! Those who know me know I’m a huge fan of @FastAPI and @pydantic , so having both involved makes it even more exciting. Looking forward to the talks, meeting people, and everything I’ll learn. Can’t wait to see what @tiangolo, @marcelotryle, and @samuelcolvin will be sharing. If you'll be around, hit me up.
    010119
  10. Pydantic@pydantic
    Why you need traces when you have logs? New blogpost by @marcelotryle
    22412.1K
  11. Pydantic@pydantic
    Pydantic
    RT @dasfacc: we improved cache reads for any @pydantic ai agent, with no code changes needed! measured on haiku 5.5: resuming a run from a StepStore: 67% → 98% right after Code Mode's dynamic catalog finds a tool: 15% → 97%
    040141
  13. Pydantic@pydantic
    Pydantic
    RT @neil_conway: Thanks for the great folks at @pydantic for sponsoring my work on Apache DataFusion and Arrow!
    010105
  14. Pydantic@pydantic
    Samuel used up his entire Fabel budget in a night, now he wants your validation that it was a good idea.
    Samuel Colvin @samuelcolvin
    Mad plot twist - what if Monty could run JS and TS code, and even run and render React components!? Just an experiment for now, but it works amazingly well. This would be especially useful if you want agents to generate UI safely. Feedback welcome.
    11543.7K
  15. Samuel Colvin✓@samuelcolvin
    Mad plot twist - what if Monty could run JS and TS code, and even run and render React components!? Just an experiment for now, but it works amazingly well. This would be especially useful if you want agents to generate UI safely. Feedback welcome.
    122606.1K
  16. Pydantic@pydantic
    Pydantic
    RT @kaxil: Already have agents? OpenAI Agents SDK, Strands, Google ADK and the Anthropic SDK get the same Airflow toolsets as the built-in @Pydantic AI operators.
    020103
  17. Pydantic@pydantic
    Pydantic
    RT @marcelotryle: Monty can pause agent-written code, turn it into bytes, and resume it later ⏸️ - dump() serializes the whole interpreter - load_snapshot() resumes it in another process or machine - Wait hours for a human approval with nothing running
    040121
  18. Pydantic@pydantic
    Last call: Tomorrow our engineers sit down with @dosu_ai on keeping internal agents’ company knowledge current: - How to evaluate an internal agent - When a person should review its work - What agents should remember, and memory drift Oct 20 at 8:30am PT/11:30am ET. Register now:
    Download
    00251.2K
  19. Pydantic@pydantic
    What's the most useful feature here? What do you want/need next?
    Samuel Colvin @samuelcolvin
    Monty v1.2.0 from @pydantic is released. Massive lease (as ever) with some very important features. Biggest features:
    21232.3K
  20. Samuel Colvin✓@samuelcolvin
    Monty v1.2.0 from @pydantic is released. Massive lease (as ever) with some very important features. Biggest features:
    Download
    55716K
  21. Pydantic@pydantic
    Pydantic
    RT @marcelotryle: Pydantic AI's CodeMode can start your tools before the model finishes writing the code ⚡ - CodeMode(speculate=['get_weather']) - A call with literal keyword arguments starts once its line streams in - In my test, 3 slow lookups went from 8.7s to 4.7s
    080125
  22. Pydantic@pydantic
    Pydantic
    RT @Glitch253: This is why reducing your dependency on corporate entities is so important. Own your own harness. Own your own hardware. Own your privacy. Why not build on top of @pydantic AI?
    Matt Johansen @mattjay
    🚨 Confirmed. OpenCode hacked. Keys stolen. Telling customers to revoke.
    040140
  23. Matt Johansen✓@mattjay
    🚨 Confirmed. OpenCode hacked. Keys stolen. Telling customers to revoke.
    Download
    1363942.3K789K
  24. Pydantic@pydantic
    As agents started opening and reviewing more of our PRs, we moved our CI to @RunsOnRunsOn. We also built a RunsOn integration to watch it in Pydantic Logfire. @RunsOnRunsOn users get dashboards for queue times, Spot interruptions, cost per job, and runner sizing.
    runs-on.com @RunsOnRunsOn
    The @pydantic team switched Logfire's CI to RunsOn on Sept 1. August, on GitHub, Depot and Ubicloud runners: $18.2k. September, on RunsOn: $5.7k, license and every AWS cost included. And they ran 31% more runner minutes in September. 🧵
    14211.8K
  25. runs-on.com@RunsOnRunsOn
    The @pydantic team switched Logfire's CI to RunsOn on Sept 1. August, on GitHub, Depot and Ubicloud runners: $18.2k. September, on RunsOn: $5.7k, license and every AWS cost included. And they ran 31% more runner minutes in September. 🧵
    1261.8K
  26. Show more posts