An open-source declarative framework for building modular AI software. Programming—not prompting—LLMs via higher-level abstractions & optimizers.
- Tweets 3,370
- Following 63
- Followers 14,349
- Likes 6,363
DSPy retweeted
DSPy methodology 🤝 System One
Program, don't prompt!
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
12
31
307
34,513
DSPy retweeted
We’re considering hosting a short, casual @DSPyOSS webinar next week to run through Jev support and design patterns. Let me know if you’re interested, and what you’d like to see covered.
14
5
86
2,872
DSPy retweeted
So excited to try this out! Love the new releases of DSPy, soooo much going on that's making me wanna use it every day
DSPy 3.4.0 out, with:
(1) native support for Jev and System One models - in the timeless DSPy syntax.
(2) a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
(3) lightning-fast import speeds for the LLM abstraction, via sibling library LM15)
1
2
21
1,810
DSPy 3.4.0 out, with:
(1) native support for Jev and System One models - in the timeless DSPy syntax.
(2) a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
(3) lightning-fast import speeds for the LLM abstraction, via sibling library LM15)
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
9
32
200
11,230
DSPy retweeted
this is an honor! 🥹
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
3
13
231
31,934
DSPy retweeted
whoa! so clean. the more the world changes, the more dspy stays the same at the interface level and very very different below the surface
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
9
13
154
11,675
DSPy retweeted
Super-cool, and a great interface through which to use Jev (if you are building stuff in Python)
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
1
10
1,388
DSPy retweeted
Just released Jev support for DSPy, with a new optimizer that calibrate’s Je’s output thresholds against your metric.
Works with existing DSPy signatures and metrics, can switch between LMs and Jev with one line.
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
4
7
91
6,092
DSPy retweeted
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
29
69
453
115,820
DSPy retweeted
Introducing ReAnchor, a DSPy optimizer built for Jev and other System One models. cmpnd.ai/blog/building-jev-p…
8
34
236
11,292
DSPy retweeted
i have something to release/share tomorrow that i think is very badly needed and has been so for a few years - it’s a tiny start and won’t be complete but i hope people will find it as useful and fun as i have enjoyed producing it
28
15
417
22,497
It's exciting that everyone's learning about "programming, rather than prompting, AI models" thanks to Jev, but the deeper point is still missing.
Asking a frontier agent is like managing a person. Hand-crafting Jev code is like hand-writing assembly for a niche RISC machine, a form of premature optimization.
Both have their use, but both are generally too extreme for engineering good systems that endure the bitter lesson.
18
29
257
12,443
this is a June 2025 talk on "Engineering AI systems that endure the bitter lesson" and we believe it might as well be a June 2026 or June 2030 talk, or indeed in 2022 when DSPy first came out:
aietalks.com/talks/on-engine…
1
5
24
1,450
DSPy retweeted
after ~3 years i think i finally now understand why the cache defaults to on in @DSPyOSS
i'm a slow learner
1
5
1,017
DSPy retweeted
hear me out: jev, but with the ability to think longer before harder decisions
oh and maybe with composite types like dicts and lists and not only bools 🤯
41
18
550
39,004
DSPy retweeted
I just tried out lm15, the new lightweight alternative to litellm from the @DSPyOSS crew.
It works with all Foundry hosted models, like Azure OpenAI, Kimi, and Anthropic.
Examples:
github.com/pamelafox/python-…
github.com/pamelafox/python-…
lm15 is way faster to import and has 0 deps!
4
3
16
2,299
