Jerry Liu(@jerryjliu0)

A lot of professional work around documents can be arbitrarily challenging to automate. It not only...

Jerry Liu (@jerryjliu0) — Mon, 19 Oct 2026 22:52:58 GMT

A lot of professional work around documents can be arbitrarily challenging to automate.

It not only requires deep visual perception/OCR around complex entities (diagrams/charts), but also requires a reasoning trace that can reconcile disparate sources of oftentimes contradictory information. 

Check out the example video below around a water loss audit. Solving this task not only requires reading the water loss diagram, but also understand the exact date and subsequent changes in the appendix.

We think hard document extraction tasks as a microcosm for professional work over PDFs. That's why we launched ExtractBenchPro - a benchmark that's challenging even for the latest frontier models (e.g. Astra/Opus). 

Even as frontier models with max thinking models push the boundaries of accuracy, there is a massive opportunity to build specialized models that can push the frontiers of cost and latency. 

For more info on ExtractBench Pro, check out our blog and the website! llamaindex.ai/blog/introduci…

Site: extractbenchpro.ai

Your browser does not support the video tag.

🔗 View on Twitter

Jerry Liu@jerryjliu0

Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction.

We see a massive long tail of document complexity in terms of use cases brought on from our customers, that are completely unsolved from a price-performance standpoint. These include trial balances, well logs, canvass binders, site plan reviews, duty reconciliations, loss runs, and more.

Most existing extraction benchmarks are saturated by the frontier models (our cheapest modes now score 93%+ on ExtractBench). So there is enormous economic value in building a complex extraction benchmark that reflects this long-tail of knowledge work.

We spent a long time curating the most complex elements that we saw from real world extraction use cases:
✅ Document conventions: containing legends or rules where the rest of the document should be interpreted
✅ Density of internal jargon: specific definitions that must be cross-referenced within a doc
✅ Aggregation and reasoning: beyond returning exact values on a page, infer values derived from aggregating multiple sources of information
✅ Drawn relations: complex diagrams like procedures and circuits
✅ Charts inside documents: read precise points within a chart
✅ Long chart sheets: some data series like well logs cram all this data into an extremely long page

Within the industries where document extraction is relevant, even small mistakes can have enormous negative impacts. Therefore our headline evaluation metric is based on perfect documents: the proportion of docs with perfect (100%) value f1 score.
(Besides this we also measure value f1 and grounded f1)

It’s an extremely hard task for even the latest frontier models to solve, topping out at ~58% or so in Value F1 score for Codex with Astra - which is quite good but extremely expensive (almost $1 per page!) Our LlamaParse agentic/agentic plus modes are 20-40% and cover the left side of the Pareto frontier, at price ranges that are feasible for full production use.

ExtractBenchPro is available today as a website/technical blog, and also on Github and HuggingFace. Check it out!

Blog: llamaindex.ai/blog/introduci…
Website: extractbenchpro.ai
Hugging Face: huggingface.co/datasets/llama…
Github (is ExtractBench): github.com/run-llama/Extr…

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬9🔄2❤️17👀1121📊10

⚡ Powered by xgo.ing

it's crazy to see how universal the acclaim for YCSE is, despite the massive divergence in opinion i...

Jerry Liu (@jerryjliu0) — Mon, 19 Oct 2026 01:57:33 GMT

it's crazy to see how universal the acclaim for YCSE is, despite the massive divergence in opinion in both interpretation as well as political affiliation.

no one can agree on anything regarding any of the people/characters in the movie, but they all agree nathan fielder produced one of the best movies of 2026.

💬3🔄0❤️11👀3374📊5

⚡ Powered by xgo.ing

One of the most important challenges in advancing enterprise AI is creating evaluation benchmarks th...

Jerry Liu (@jerryjliu0) — Sun, 18 Oct 2026 18:26:33 GMT

One of the most important challenges in advancing enterprise AI is creating evaluation benchmarks that reflect their real data and workflows - without using their actual data or workflows due to privacy/security restrictions 

We see our customers upload a massive long tail of complex documents for extraction, which break even the most complex frontier models. 

In order to create a representative benchmark that adheres to this complexity, we invested heavily in a synthetic pipeline that can mine real document templates, but allow flexibility for variation of not only the data, but the template and rendering.

Come check out ExtractBench Pro! llamaindex.ai/blog/introduci…

Website: extractbenchpro.ai

Your browser does not support the video tag.

🔗 View on Twitter

Jerry Liu@jerryjliu0

Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction.

We see a massive long tail of document complexity in terms of use cases brought on from our customers, that are completely unsolved from a price-performance standpoint. These include trial balances, well logs, canvass binders, site plan reviews, duty reconciliations, loss runs, and more.

Most existing extraction benchmarks are saturated by the frontier models (our cheapest modes now score 93%+ on ExtractBench). So there is enormous economic value in building a complex extraction benchmark that reflects this long-tail of knowledge work.

We spent a long time curating the most complex elements that we saw from real world extraction use cases:
✅ Document conventions: containing legends or rules where the rest of the document should be interpreted
✅ Density of internal jargon: specific definitions that must be cross-referenced within a doc
✅ Aggregation and reasoning: beyond returning exact values on a page, infer values derived from aggregating multiple sources of information
✅ Drawn relations: complex diagrams like procedures and circuits
✅ Charts inside documents: read precise points within a chart
✅ Long chart sheets: some data series like well logs cram all this data into an extremely long page

Within the industries where document extraction is relevant, even small mistakes can have enormous negative impacts. Therefore our headline evaluation metric is based on perfect documents: the proportion of docs with perfect (100%) value f1 score.
(Besides this we also measure value f1 and grounded f1)

It’s an extremely hard task for even the latest frontier models to solve, topping out at ~58% or so in Value F1 score for Codex with Astra - which is quite good but extremely expensive (almost $1 per page!) Our LlamaParse agentic/agentic plus modes are 20-40% and cover the left side of the Pareto frontier, at price ranges that are feasible for full production use.

ExtractBenchPro is available today as a website/technical blog, and also on Github and HuggingFace. Check it out!

Blog: llamaindex.ai/blog/introduci…
Website: extractbenchpro.ai
Hugging Face: huggingface.co/datasets/llama…
Github (is ExtractBench): github.com/run-llama/Extr…

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬29🔄4❤️37👀3637📊31

⚡ Powered by xgo.ing

All I can say after watching it is, I am so excited for The Rehearsal season 3

Jerry Liu (@jerryjliu0) — Sun, 18 Oct 2026 05:57:39 GMT

All I can say after watching it is,

I am so excited for The Rehearsal season 3

Jerry Liu@jerryjliu0

please don't post spoilers on the movie 🙏🙏 

at least until sunday lol

🔗 View Quoted Tweet

💬3🔄0❤️13👀4165📊4

⚡ Powered by xgo.ing

There’s a crazy number of complicated PDFs out there. Each PDF is a self-contained environment for ...

Jerry Liu (@jerryjliu0) — Sun, 18 Oct 2026 00:56:58 GMT

There’s a crazy number of complicated PDFs out there.

Each PDF is a self-contained environment for knowledge work, and it can take humans hours to read, analyze, and synthesize a document to solve their task.

@llama_index is a research lab dedicated to pushing the frontier of document work. ExtractBench Pro represents our latest effort to extract from the hardest documents at the frontier of accuracy and cost. Even as frontier intelligence starts to do anything with enough compute, there are massive opportunities to solve any document work extremely quickly and quickly.

If you haven’t already, check out ExtractBench Pro: llamaindex.ai/blog/introduci…

If you’re looking to migrate off of legacy solutions, come talk to us! llamaindex.ai/contact

Your browser does not support the video tag.

🔗 View on Twitter

Jerry Liu@jerryjliu0

Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction.

We see a massive long tail of document complexity in terms of use cases brought on from our customers, that are completely unsolved from a price-performance standpoint. These include trial balances, well logs, canvass binders, site plan reviews, duty reconciliations, loss runs, and more.

Most existing extraction benchmarks are saturated by the frontier models (our cheapest modes now score 93%+ on ExtractBench). So there is enormous economic value in building a complex extraction benchmark that reflects this long-tail of knowledge work.

We spent a long time curating the most complex elements that we saw from real world extraction use cases:
✅ Document conventions: containing legends or rules where the rest of the document should be interpreted
✅ Density of internal jargon: specific definitions that must be cross-referenced within a doc
✅ Aggregation and reasoning: beyond returning exact values on a page, infer values derived from aggregating multiple sources of information
✅ Drawn relations: complex diagrams like procedures and circuits
✅ Charts inside documents: read precise points within a chart
✅ Long chart sheets: some data series like well logs cram all this data into an extremely long page

Within the industries where document extraction is relevant, even small mistakes can have enormous negative impacts. Therefore our headline evaluation metric is based on perfect documents: the proportion of docs with perfect (100%) value f1 score.
(Besides this we also measure value f1 and grounded f1)

It’s an extremely hard task for even the latest frontier models to solve, topping out at ~58% or so in Value F1 score for Codex with Astra - which is quite good but extremely expensive (almost $1 per page!) Our LlamaParse agentic/agentic plus modes are 20-40% and cover the left side of the Pareto frontier, at price ranges that are feasible for full production use.

ExtractBenchPro is available today as a website/technical blog, and also on Github and HuggingFace. Check it out!

Blog: llamaindex.ai/blog/introduci…
Website: extractbenchpro.ai
Hugging Face: huggingface.co/datasets/llama…
Github (is ExtractBench): github.com/run-llama/Extr…

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬29🔄5❤️88👀5563📊37

⚡ Powered by xgo.ing

my guy is running gradient descent on math

Jerry Liu (@jerryjliu0) — Sun, 18 Oct 2026 00:37:37 GMT

my guy is running gradient descent on math

Rohan Arun@RohanArun

From κ = 2⁻¹⁸² to 0.001734815770 :)

Integer Multiplication community auto-research highlights since 10/15 (+1.23%):

1) The big headline of the day is that a free model, @opencode's Space Bunny Max, just set the record for the first time, 4 times in a row!

As far as I know, this is the first time a free model has submitted a best bound for Integer Multiplication.

Expensive Max subscriptions previously kept some of our best contributors and experts out, but the barrier of entry for frontier auto-research dropped to 0 :)

Claude Opus 5.5 helped verify and push the PRs over the line. I still recommend having a frontier model verify, but other contributors can do that. 

Progress was slowing over the last day (relative to one new κ per minute, which is still wild for math research), but Space Bunny Max just injected fresh ideas that boosted our PR rate back to one per minute.

2) A community Lean proof of the original OpenAI Integer Multiplication paper has been completed!

@platypii et. al completed the proof in 6 days, and @0xdoug's repo is about 10–20% verified now.

To be clear, most of our work still needs verification but the Lean proofs are catching up :)

3) 41–51% of PRs now mention a prior research-only submission, up from 0–3% before @0xdoug's autoresearch.md prompt.

His prompt guides new contributors to specific research tracks rather than incremental updates.

We're moving away from "brute force research" towards more coordinated efforts across an army of agents.

If you're contributing research PRs, even negative results without a new record, you're meaningfully helping us and I appreciate it!

Interesting observations:

1) The results from free models are related to the recent "local minima" solutions in context engineering that have been popping up:

Forgetting context to "refresh" by @ShangyinT.

Branching into different solutions by @mao_yuzhen.

Fork and flush by @DimitrisPapail.

Now even free models contribute to escaping local minima by exploring new solutions, since they're trained on different data.

There's likely a lot of room for improvement now in context engineering, consensus across different models, and collaboration strategies, which we will test next.

2) Frontier research as a benchmark helped us find a successful context-sharing strategy for parallel agents.

Basically, we can rewind frontier research GitHub repos to an earlier branch and benchmark agent harnesses, prompts, context designs, models, etc.

With this benchmark, we found that you can distill constantly to local files, then periodically concatenate them instantly into parallel agent sessions to save 10–20% of the time it takes to reach the next breakthrough.

3) There's a Codex reset by EOD PST, so you can copy this prompt to run unused tokens on frontier research :)

"""
Contribute one useful, independently reviewable result to github.com/CrocSwap/integ…. Read AGENTS.md, autoresearch.md and CONTRIBUTING.md from the current main branch first. Use the repository's task-selection and overlap process unless I assigned a specific task. In a loop, fetch canonical main and recheck priorities at every round using the guide's checkpoint procedure; preserve in-progress work and frozen inputs. Work for one session of about 60 minutes unless I specify another budget, keeping the final quarter for verification and a concise handoff, and report the exact scope, evidence, remaining obligations and predecessor credit. You may post routine claims, questions, findings, handoffs and feedback in Discussions and open a focused PR without asking me again for each action. Do not merge, close others' PRs or change the selected frontier.
"""
You can also copy this prompt into OpenCode now!

Now that anyone can contribute for free to frontier auto-research, repost this and sharing it with friends. 🙏

4) PR stats:

850 PRs from 67 contributors since 10/15, and 21 of those contributors are first-timers.

55 PRs claim a new conditional κ in the title.

Codex/OpenAI in 480 PRs (56%), Claude in 147 (17%), the OpenCode harness in 60 (7%), Cursor in 16.

Most prolific contributors by PR count: petalas: 164, ywz65: 147, navinor: 49, commdata2338: 44, RossRichardson: 43, Arham-Tahir64: 39, thakrarsagar: 34, VictorLiwentao: 30, tekuuu: 28, DreamingOfClouds: 23, jmbarrancoml: 21, jeremykothe2: 21, bitplane: 13, Gonzalo-Mier: 12, Dugongue: 12, mdumitrean: 11, theon199: 9, jwr404: 9, jamesyc: 9, kumagi: 8, CubeJerry: 8, rohanarun: 6, jeremydiviney: 6, dcharb78: 6, 1337isnot1337: 6, starship-s: 6, AAAmirinu: 6.

Conditional κ = 0.001734815770385098563484.

Just submitted a new best bound :)
github.com/CrocSwap/integ…d

🔗 View Quoted Tweet

💬3🔄1❤️12👀3233📊5

⚡ Powered by xgo.ing

please don't post spoilers on the movie 🙏🙏 at least until sunday lol

Jerry Liu (@jerryjliu0) — Sat, 17 Oct 2026 02:35:21 GMT

please don't post spoilers on the movie 🙏🙏 

at least until sunday lol

💬6🔄0❤️19👀15223📊9

⚡ Powered by xgo.ing

A lot of real-world documents embed extremely dense charts that require humans to not only "perceive...

Jerry Liu (@jerryjliu0) — Fri, 16 Oct 2026 22:45:10 GMT

A lot of real-world documents embed extremely dense charts that require humans to not only "perceive" the chart values, but also reason over them and reconcile with other complex elements (e.g. tables) to solve a knowledge task.

We've encoded this complex chart+document reasoning within our ExtractBench Pro benchmark, released yesterday. Even the best frontier models struggle at solving complex extraction tasks perfectly over these documents.

For more details on ExtractBench Pro, check out the blog: llamaindex.ai/blog/introduci…

If you have complex extraction needs at scale, check out LlamaParse: cloud.llamaindex.ai

Jerry Liu@jerryjliu0

Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction.

We see a massive long tail of document complexity in terms of use cases brought on from our customers, that are completely unsolved from a price-performance standpoint. These include trial balances, well logs, canvass binders, site plan reviews, duty reconciliations, loss runs, and more.

Most existing extraction benchmarks are saturated by the frontier models (our cheapest modes now score 93%+ on ExtractBench). So there is enormous economic value in building a complex extraction benchmark that reflects this long-tail of knowledge work.

We spent a long time curating the most complex elements that we saw from real world extraction use cases:
✅ Document conventions: containing legends or rules where the rest of the document should be interpreted
✅ Density of internal jargon: specific definitions that must be cross-referenced within a doc
✅ Aggregation and reasoning: beyond returning exact values on a page, infer values derived from aggregating multiple sources of information
✅ Drawn relations: complex diagrams like procedures and circuits
✅ Charts inside documents: read precise points within a chart
✅ Long chart sheets: some data series like well logs cram all this data into an extremely long page

Within the industries where document extraction is relevant, even small mistakes can have enormous negative impacts. Therefore our headline evaluation metric is based on perfect documents: the proportion of docs with perfect (100%) value f1 score.
(Besides this we also measure value f1 and grounded f1)

It’s an extremely hard task for even the latest frontier models to solve, topping out at ~58% or so in Value F1 score for Codex with Astra - which is quite good but extremely expensive (almost $1 per page!) Our LlamaParse agentic/agentic plus modes are 20-40% and cover the left side of the Pareto frontier, at price ranges that are feasible for full production use.

ExtractBenchPro is available today as a website/technical blog, and also on Github and HuggingFace. Check it out!

Blog: llamaindex.ai/blog/introduci…
Website: extractbenchpro.ai
Hugging Face: huggingface.co/datasets/llama…
Github (is ExtractBench): github.com/run-llama/Extr…

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬22🔄5❤️69👀4599📊28

⚡ Powered by xgo.ing

congrats on the launch!!

Jerry Liu (@jerryjliu0) — Fri, 16 Oct 2026 04:00:26 GMT

congrats on the launch!!

Zakaria@zakariaornot

Introducing overshoot,  it will cut your GPU bill in half with zero quality loss, or you don’t pay.

Over the past few months, we have forward deployed into AI teams to optimize their inference stacks. They have now cumulatively saved over $5M.

Overshoot is software that bakes in our hard-earned inference learnings into a scalable solution.

overshoot.ai

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬3🔄1❤️10👀3114📊4

⚡ Powered by xgo.ing

irl rocket league

Jerry Liu (@jerryjliu0) — Fri, 16 Oct 2026 03:38:24 GMT

irl rocket league

Tesla@Tesla

Real life, baby

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬4🔄1❤️30👀4514📊6

⚡ Powered by xgo.ing

thought this was AI wtf

Jerry Liu (@jerryjliu0) — Fri, 16 Oct 2026 02:39:16 GMT

thought this was AI wtf

Elon Musk@elonmusk

Turns out the future does have flying cars

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬1🔄1❤️32👀7711📊4

⚡ Powered by xgo.ing

claude cowork really needs to let you fork a conversation this feature exists in other apps (e.g. c...

Jerry Liu (@jerryjliu0) — Fri, 16 Oct 2026 01:22:26 GMT

claude cowork really needs to let you fork a conversation

this feature exists in other apps (e.g. codex app, claude code cli). it's one of the main reasons i've been using codex as a daily driver for the past few months

💬44🔄0❤️42👀5901📊51

⚡ Powered by xgo.ing

Document extraction is an arbitrarily complex reasoning problem for both frontier models and humans....

Jerry Liu (@jerryjliu0) — Thu, 15 Oct 2026 23:08:52 GMT

Document extraction is an arbitrarily complex reasoning problem for both frontier models and humans.

✅ The schema encodes essentially hundreds of interrelated reasoning tasks
✅ A PDF can be hundreds or even thousands of pages - it’s an open-world information environment for LLMs to explore (similar to why gdp.pdf is hard)
✅ Solving even a single task is a hard reasoning problem over this environment, where the right dependencies must be reasoned over
✅ The information can be scattered across extraordinarily complex dimensions, from sections, tables, charts, diagrams, handwriting, and more.

Document work is a microcosm of knowledge work as a whole. We’re invested in developing the hardest benchmarks that reflect real-world professional work, challenge the capabilities of frontier models, and incentivize specialized model development for document work.

We released ExtractBench Pro today. Check out the below example as an example of a reasoning task that requires finding the right constraints, then pulling information from tables. llamaindex.ai/blog/introduci…

ExtractBench Pro: extractbenchpro.ai

Jerry Liu@jerryjliu0

Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction.

We see a massive long tail of document complexity in terms of use cases brought on from our customers, that are completely unsolved from a price-performance standpoint. These include trial balances, well logs, canvass binders, site plan reviews, duty reconciliations, loss runs, and more.

Most existing extraction benchmarks are saturated by the frontier models (our cheapest modes now score 93%+ on ExtractBench). So there is enormous economic value in building a complex extraction benchmark that reflects this long-tail of knowledge work.

We spent a long time curating the most complex elements that we saw from real world extraction use cases:
✅ Document conventions: containing legends or rules where the rest of the document should be interpreted
✅ Density of internal jargon: specific definitions that must be cross-referenced within a doc
✅ Aggregation and reasoning: beyond returning exact values on a page, infer values derived from aggregating multiple sources of information
✅ Drawn relations: complex diagrams like procedures and circuits
✅ Charts inside documents: read precise points within a chart
✅ Long chart sheets: some data series like well logs cram all this data into an extremely long page

Within the industries where document extraction is relevant, even small mistakes can have enormous negative impacts. Therefore our headline evaluation metric is based on perfect documents: the proportion of docs with perfect (100%) value f1 score.
(Besides this we also measure value f1 and grounded f1)

It’s an extremely hard task for even the latest frontier models to solve, topping out at ~58% or so in Value F1 score for Codex with Astra - which is quite good but extremely expensive (almost $1 per page!) Our LlamaParse agentic/agentic plus modes are 20-40% and cover the left side of the Pareto frontier, at price ranges that are feasible for full production use.

ExtractBenchPro is available today as a website/technical blog, and also on Github and HuggingFace. Check it out!

Blog: llamaindex.ai/blog/introduci…
Website: extractbenchpro.ai
Hugging Face: huggingface.co/datasets/llama…
Github (is ExtractBench): github.com/run-llama/Extr…

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬15🔄7❤️59👀5779📊28

⚡ Powered by xgo.ing

Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction. We ...

Jerry Liu (@jerryjliu0) — Thu, 15 Oct 2026 16:16:18 GMT

Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction.

We see a massive long tail of document complexity in terms of use cases brought on from our customers, that are completely unsolved from a price-performance standpoint. These include trial balances, well logs, canvass binders, site plan reviews, duty reconciliations, loss runs, and more.

Most existing extraction benchmarks are saturated by the frontier models (our cheapest modes now score 93%+ on ExtractBench). So there is enormous economic value in building a complex extraction benchmark that reflects this long-tail of knowledge work.

We spent a long time curating the most complex elements that we saw from real world extraction use cases:
✅ Document conventions: containing legends or rules where the rest of the document should be interpreted
✅ Density of internal jargon: specific definitions that must be cross-referenced within a doc
✅ Aggregation and reasoning: beyond returning exact values on a page, infer values derived from aggregating multiple sources of information
✅ Drawn relations: complex diagrams like procedures and circuits
✅ Charts inside documents: read precise points within a chart
✅ Long chart sheets: some data series like well logs cram all this data into an extremely long page

Within the industries where document extraction is relevant, even small mistakes can have enormous negative impacts. Therefore our headline evaluation metric is based on perfect documents: the proportion of docs with perfect (100%) value f1 score.
(Besides this we also measure value f1 and grounded f1)

It’s an extremely hard task for even the latest frontier models to solve, topping out at ~58% or so in Value F1 score for Codex with Astra - which is quite good but extremely expensive (almost $1 per page!) Our LlamaParse agentic/agentic plus modes are 20-40% and cover the left side of the Pareto frontier, at price ranges that are feasible for full production use.

ExtractBenchPro is available today as a website/technical blog, and also on Github and HuggingFace. Check it out!

Blog: llamaindex.ai/blog/introduci…
Website: extractbenchpro.ai
Hugging Face: huggingface.co/datasets/llama…
Github (is ExtractBench): github.com/run-llama/Extr…

Your browser does not support the video tag.

🔗 View on Twitter

💬38🔄18❤️130👀27515📊74

⚡ Powered by xgo.ing

this is so good

Jerry Liu (@jerryjliu0) — Wed, 14 Oct 2026 02:49:01 GMT

this is so good

Alexandr Wang@alexandr_wang

you cannot out-shitpost me

(with the help of muse wonder full)

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬0🔄0❤️20👀10875📊3

⚡ Powered by xgo.ing

Gemini 4 Argon, Muse Spark 1.3, and Muse Wonder 1.0 are all really good at pdf parsing. Come check ...

Jerry Liu (@jerryjliu0) — Wed, 14 Oct 2026 02:40:32 GMT

Gemini 4 Argon, Muse Spark 1.3, and Muse Wonder 1.0 are all really good at pdf parsing.

Come check it out on OpenDocRouter! opendocrouter.ai

Your browser does not support the video tag.

🔗 View on Twitter

LlamaIndex 🦙@llama_index

OpenDocRouter makes it dead-simple to start parsing your documents across on ton of frontier Models (gemini, luna) and latest OSS models (LightOnOCR-3, MinerU2.5-pro). Parse your first document in 60 seconds!

opendocrouter.ai

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬2🔄2❤️19👀3266📊5

⚡ Powered by xgo.ing

Benchmarks are becoming the core competency of any company. Our company mission is to build the wor...

Jerry Liu (@jerryjliu0) — Wed, 14 Oct 2026 02:37:56 GMT

Benchmarks are becoming the core competency of any company.

Our company mission is to build the world’s best models for document processing. This also means that we have to build the world’s best benchmarks for doc parsing and extraction.

It’s been incredible to see the adoption of ParseBench the past few months. We have 26k+ monthly downloads on Hugging Face, 147+ models benchmarked, and has been cited by other labs in the space (Cohere, Mistral, etc.).

We use this as a benchmark to push our own internal capabilities within LlamaParse, and also to track the advancement of recent frontier/open-weight models (most recently, Gemini 4 Argon and Muse Spark/Wonder).

We’re happy to help apply this to any company looking at evaluating document processing solutions. Even if you don’t use our docs, you may find our metrics for evaluating various dimensions (e.g. tables, reading order, layout) useful.

Check out ParseBench: parsebench.ai

Hugging Face: huggingface.co/datasets/llama…

💬43🔄8❤️64👀5082📊56

⚡ Powered by xgo.ing

doc chai is really good spiced, creamy, refreshing. we went through like 10 variations of this bef...

Jerry Liu (@jerryjliu0) — Wed, 14 Oct 2026 02:29:28 GMT

doc chai is really good 

spiced, creamy, refreshing. we went through like 10 variations of this before settling on this one 

come try it!!

Imaan Sultan 🇵🇰🇸🇦@imaanxsultan

hi twitter friends

im doing a party on thursday and i'd love for you to come! 

@llama_index is launching its first drink at @UseCorgi cafe and i got to work on designing it 🥹 we have a dj and free drinks for 100 ppl from 8-11 am

would love for you to try + give me your review, link to rsvp is in the comments :D

🔗 View Quoted Tweet

💬3🔄0❤️3👀2864📊3

⚡ Powered by xgo.ing

We benchmarked Gemini 4 Argon, Muse Wonder, and Muse Spark comprehensively on hard document OCR task...

Jerry Liu (@jerryjliu0) — Tue, 13 Oct 2026 21:15:17 GMT

We benchmarked Gemini 4 Argon, Muse Wonder, and Muse Spark comprehensively on hard document OCR tasks.

They're all pretty good! 
- Muse Spark and Gemini 4 Argon* are SOTA at their respective price points for table understanding among frontier models. 
- Wonder and Argon are also quite good at chart understanding, reading order, and formatting recognition

It's exciting that frontier models are starting to ramp up on visual understanding capabillities. 

We're committed to open benchmarking of all available models even as we aim to build the best specialized models for document parsing and extraction. 

You can check out the full set of models on OpenDocRouter: opendocrouter.ai

If you want a dedicated OCR solution without choosing, come check out LlamaParse: cloud.llamaindex.ai

Logan Markewich@LoganMarkewich

Added a few new models to OpenDocRouter (gemini-4-argon, Muse Spark/Wonder).

Interesting to see the latency for Muse Wonder so low. Gemini's costs are low but their introductory rates mean the price will double later this year.

🔗 View Quoted Tweet

💬18🔄6❤️84👀33037📊29

⚡ Powered by xgo.ing

We launched a drink at @UseCorgi cafe 🐶 Doc Chai: - iced toasted brown sugar chai - whole spices ...

Jerry Liu (@jerryjliu0) — Tue, 13 Oct 2026 20:21:25 GMT

We launched a drink at @UseCorgi cafe 🐶

Doc Chai: 
- iced toasted brown sugar chai
- whole spices and black tea
- your choice of milk 

approved by me, @imaanxsultan, and a bunch of other folks on the @llama_index team. 10/10 would recommend 

Come check out our launch party on Thursday!luma.com/llamai-68eag

LlamaIndex 🦙@llama_index

Doc Chai is real, and on Thursday you can drink it ☕  

We’re throwing a launch party for our first drink with @useCorgi Cafe: an iced toasted brown sugar chai!  

For many of us on the team, chai is what home tastes like.  

Come say hi, 600+ people have RSVP'ed and the first 100 drinks are on us. We'll have a live DJ to wake everyone up as well :)  

8-11 AM, Corgi Cafe, FiDi 🦙  

RSVP link in the comments below.

🔗 View Quoted Tweet

💬12🔄4❤️35👀3547📊14

⚡ Powered by xgo.ing

you guys are really living life at the edge

Jerry Liu (@jerryjliu0) — Tue, 13 Oct 2026 05:41:51 GMT

you guys are really living life at the edge

Deedy@deedydas

There is no faster city for flying than San Francisco.

10pm: Leave house
32pm: Arrive at SFO
40pm: Get to gate
43pm: Boarding starts
48pm: At seat

I’ve never done a 38min house-seat time before.

🔗 View Quoted Tweet

💬3🔄1❤️190👀284071📊23

⚡ Powered by xgo.ing

least clickbait slop post

Jerry Liu (@jerryjliu0) — Mon, 12 Oct 2026 23:56:59 GMT

least clickbait slop post

Divyanshi@Divyyanshishrma

China has killed the entire vector database industry.

They open-sourced TencentDB Agent Memory. It gives any AI agent long-term memory that runs 100% locally.

No Pinecone. No cloud APIs. No repeating yourself every session.

- 61% fewer tokens
- PersonaMem accuracy: 48% → 76%
- Zero external API dependencies
- Runs on plain SQLite

Most memory systems compress your history into an opaque vector pile. when recall goes wrong, you're guessing. this one doesn't compress, it builds a semantic pyramid.

L0 Conversation → L1 Atom → L2 Scenario → L3 Persona.

Short-term state gets encoded as a Mermaid graph in your agent's context. verbose tool logs get offloaded to disk. when the agent needs proof, it drills back via node_id to the exact raw log.

no lossy compression. every layer is readable markdown you can just open and inspect.

5.1k stars. 100% Open Source.

🔗 View Quoted Tweet

💬8🔄2❤️55👀11089📊23

⚡ Powered by xgo.ing

We hosted an awesome black tie gala last Friday to close out SF tech week huge shoutout to everyon...

Jerry Liu (@jerryjliu0) — Mon, 12 Oct 2026 18:23:56 GMT

We hosted an awesome black tie gala last Friday to close out SF tech week 

huge shoutout to everyone involved with organizing! @ValyuOfficial@MongoDB@daytonaio@arceuslegal@obvious@thelinqapp@lemma_@Tiny_Fish + ofc @imaanxsultan from our side

LlamaIndex 🦙@llama_index

Three events, one award, a very respectable 4th place in hot sauce.

Our SF Tech Week:

→ AI Demo Night w/ @ClickHouse + @artie_labs: @imaanxsultan won best demo

→ @matissemlk took 4th in the Blazel hot sauce competition

→ Co-hosted the biggest night of SF Tech Week with @SpaceX, @Lemma_, @Daytonaio, @Tiny_Fish, and more! Agent Economy Gala: 500+ founders + investors in suits and gowns

Thanks to everyone who joined us for a very busy, but fun SF Tech Week :)

🔗 View Quoted Tweet

💬12🔄4❤️29👀3323📊13

⚡ Powered by xgo.ing

Unlimited OCR is a good example of a low-parameter OCR model that can process any document extremely...

Jerry Liu (@jerryjliu0) — Sun, 11 Oct 2026 21:12:01 GMT

Unlimited OCR is a good example of a low-parameter OCR model that can process any document extremely quickly.

It's good at reading order and layout, and decent over tables. It is slightly better than PaddleOCR on tables and slightly worse on layout.

There are tradeoffs though - it ignores visual elements, formatting, and the most complex tables. It scores ~46.2% and ranks #115 on ParseBench. 

There's plenty of opportunities to improve this frontier, but Unlimited OCR is still a good milestone! 

Check out ParseBench: parsebench.ai

Brooks Whale X 🐋@BrooksWhaleX

🚨 CHINA JUST OPEN SOURCED AN OCR THAT EATS 100 PAGE PDFS IN ONE PASS

It’s called Unlimited OCR. Only 3B params. Runs locally.

Every other OCR tool chops your doc into pages and loses the thread. This one reads the whole thing in one continuous pass.

→ One shot full context parsing (32K context window)
→ Multilingual out of the box
→ 93% on the standard parsing benchmark (6 points over baseline)
→ Under 0.11 error rate past 40 pages
→ Runs 100% locally on your own hardware
→ Works with Transformers, vLLM, SGLang, Docker, Ollama, llama.cpp

Traditional cloud OCR (Textract, Google Vision, Azure Doc Intelligence) costs $1.50 to $15 per 1,000 pages.

This runs on your machine. For free. Forever.

Baidu built it explicitly to push DeepSeek OCR one step further. Already at 1.9M downloads on Hugging Face and most people have no idea it exists yet.

100% open source.

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬26🔄37❤️323👀18269📊104

⚡ Powered by xgo.ing

One of the biggest historical bottlenecks for training specialized models has been curating ground-t...

Jerry Liu (@jerryjliu0) — Sun, 11 Oct 2026 16:24:44 GMT

One of the biggest historical bottlenecks for training specialized models has been curating ground-truth. This is typically a very labor intensive process done with human supervision.

Now frontier models/agents are getting better and better at bootstrapping ground-truth from scratch, especially for verifiable tasks. They can do this either in a single-shot setting for simple tasks or simulate longer rollouts with rewards for longer tasks. 

This paves the way for anyone to build mostly automated research loops without needing deep ML expertise or deep access to human labor for data curation. That part can be offloaded to the frontier model/agent; the human's main goal is to define the right evals. This is especially useful for anyone who wants to optimize for a given task at the price-performance frontier. Posttraining open-weight models/harness should be made available to everyone, not just labs/specific startups.

Flow for the average user, who is only responsible for the first step:
1. define goal/evals for a given task
2. run frontier agents to gather ground-truth signals for the task. they can bootstrap annotations directly, and reach out to human review where needed 
3. train specialized models (and harness) against this task

💬51🔄12❤️139👀11083📊85

⚡ Powered by xgo.ing

linea (!)

Jerry Liu (@jerryjliu0) — Sun, 11 Oct 2026 15:47:31 GMT

linea (!)

Lee Edwards@ledwards

@AstasiaMyers One slight modification.

🔗 View Quoted Tweet

💬2🔄0❤️21👀4964📊5

⚡ Powered by xgo.ing

long live RAG

Jerry Liu (@jerryjliu0) — Sat, 10 Oct 2026 19:08:59 GMT

long live RAG

Pope Leo XIV@Pontifex

Intelligence must draw on memory. The mind must not simply compile data — as an algorithm now does more quickly than we can — but recall lived experiences, which contain depths of meaning that only the human soul can recognize, and connect more deeply with the goodness of these experiences.  Everything that is good begins “with the Spirit” and thus possesses a breath, a dynamism and a sense of communion that, once experienced, are engraved upon the heart. vatican.va/content/leo-xi…

🔗 View Quoted Tweet

💬5🔄3❤️65👀9648📊14

⚡ Powered by xgo.ing

Mistral OCR is on OpenDocRouter! For its price, it is decent over tables and reading order. It's a...

Jerry Liu (@jerryjliu0) — Sat, 10 Oct 2026 19:02:10 GMT

Mistral OCR is on OpenDocRouter! 

For its price, it is decent over tables and reading order. It's also blazing fast at p50 1.8s per page 

We're constantly bringing in new models since OpenDocRouter launched last week: opendocrouter.ai

LlamaIndex 🦙@llama_index

Today we're announcing OpenDocRouter: every model for document parsing under one API.

There are ~4,000 OCR models on Hugging Face, and the frontier labs ship a new one nearly every month. Whatever's best for your docs today won't be by Q1.

So stop asking which model to use for doc parsing. Ask how fast you can switch.

✅ Switch models in one line: same request, same markdown output, no new prompts or integrations
✅ Any model you want: 10 frontier and open-source models at launch, including Claude Opus 5.5, Gemini 3.8 Flash, GPT-6 Luna, MinerU2.5-Pro and PaddleOCR-VL-1.6
✅ Choose with receipts: every model scored on ParseBench for quality and cost
✅ Traceable output from any model: "layout: true" adds grounded bounding boxes and the same layout classes, even for models that don't support it natively
✅ Pay only for what works: per-token pricing, failed pages never charged, top up from $25

The spread is the point: $0.86 to $48.82 per 1,000 pages, depending on what your documents actually need.

Live now → opendocrouter.ai

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬19🔄7❤️79👀6627📊28

⚡ Powered by xgo.ing

"a corollary is that information about product quality will become more important and impactful" T...

Jerry Liu (@jerryjliu0) — Sat, 10 Oct 2026 18:11:41 GMT

"a corollary is that information about product quality will become more important and impactful" 

This feels like the next evolution of SEO/AEO that every product will try to game. Even if agents do the equivalent of ~10 hours of research on a product, it's less important on whether the product is actually better vs. the agent's perception that the product is better.

Agents might do deep research on evals and benchmarks; yet anyone who's worked on evals knows that it's hard to create a truly unbiased one that can universally measure product quality. They might optimize for human preferences, in which case companies will optimize for user reviews and feedback. 

Companies will optimize for whatever preferences the top AI models have been posttrained on.

Patrick Collison@patrickc

Assorted things I've been mulling over regarding agents in the economy:

• Companies are in some sense the original superintelligence. It's often the case that they can use their amortized cognitive surplus to advance their interests against those of certain consumers: relative to the optimization capacity of companies, consumers can be inattentive, underinformed, irrational, etc. (Think, say, of some of the classic insurance upsells at the end of a booking process, which is typically priced substantially above what a regular policy would cost.) Herbert Simon introduced the idea of bounded rationality, and firms often end up benefiting from it. Thanks to personal agents, we will probably get somewhat more rational in how we engage with companies.

• How exactly will outcomes change? Well, certain kinds of price discrimination will work less well (how do coupons work if every agent hunts fastidiously for them?), and certain kinds of customer cross-subsidy will change (today, people who forget to cancel subscriptions subsidize those who don't). Taken across the economy as a whole, I'm not sure what the incidence of all of this will be.

• In corporate strategy, people have long debated the relative importance of distribution and product quality. Perhaps you've made something better, but does anyone know about it? Won't people just continue to buy the ACME Corp Mousetraps? Personal agents will, I think, act as a kind of structural subsidy for product quality. How would markets look different if individuals always spent at least 10 hours researching their purchases? On the margin, focusing on making a product better is, I think, going to become a more effective strategy for companies. This is good!

• It true, a corollary is that information about product quality will become more important and impactful. Today, most people have very limited ability to aggregate and integrate such data, but agents will have insatiable thirst for such signals. (It's often said that data is becoming more valuable because it helps in the training of models. That's obviously true, but I think there's a secondary way in which it's more valuable simply as a complement to cheaper intelligence.)

• A big question is whether agents will increase superstar effects (rational agents all settle on the same product) or increase dispersion (because everyone has slightly different tastes and preferences, which agents are good at eliciting, finally overcoming the artificial clustering we've been subjected to).

• Many internet companies have monetized demand routing, which is in turn often predicated on short attention spans. "I'll click on the first thing I see." This has led to some amount of inefficiency, as companies try to propitiate capricious ranking algorithms and incur various placement fees. It's not all deadweight loss (willingness to pay is itself a kind of signal, which is valuable to buyers), but it's not great. How will the nature and value of this routing change in a world of personal agents? I don't know the answer, but I think that the effects might be pretty big.

• Much of what I'm writing hinges on the assumption that personal agents will be on the consumer's side. I suspect that that will be the winning strategy, but whether it's true, and what it means to be true, involves a lot of other questions.

Taken as a whole, the market is a massively parallelized, decentralized, and highly imperfect value attribution machine, and, taking stock of all of the effects, it seems to me that the quality point may well be the biggest: that agents will change the nature of the backpropagation such that the rewards for better products increase. Maybe this is too optimistic, but I suspect (and hope) that companies that make things that they're proud of will feel that the universe is a little more partisan in their favor.

🔗 View Quoted Tweet

💬22🔄1❤️14👀3507📊20

⚡ Powered by xgo.ing

These days, you can pretty much solve any task by defining an eval and hillclimbing over it, instead...

Jerry Liu (@jerryjliu0) — Fri, 09 Oct 2026 05:58:08 GMT

These days, you can pretty much solve any task by defining an eval and hillclimbing over it, instead of directly defining the deterministic/agentic workflow to solve it.

The data provider companies' entire job is to define evals for all economic activity to make the frontier models capable of doing anything. 

Your job then becomes pointing frontier intelligence in the right direction - defining what to solve, and how to measure what good looks like. 

Agent application interfaces will evolve to capture this. The most complex processes will still need some sort of explicit workflow builder interface, but most tasks can be compressed into goals and eval instructions.

💬116🔄82❤️1136👀75420📊468

⚡ Powered by xgo.ing

Today we’re announcing dots https://t.co/nkYyYDtlnL

Jerry Liu (@jerryjliu0) — Fri, 09 Oct 2026 02:30:44 GMT

Today we’re announcing dots

dot.com

Tibo@thsottiaux

Today, we are announcing ChatGPT.

It is here: chatgpt.com

🔗 View Quoted Tweet

💬19🔄3❤️267👀88734📊44

⚡ Powered by xgo.ing

i'm pulling out my suit for this Huge shoutout to @imaanxsultan and everyone else for organizing!

Jerry Liu (@jerryjliu0) — Fri, 09 Oct 2026 01:02:52 GMT

i'm pulling out my suit for this 

Huge shoutout to @imaanxsultan and everyone else for organizing!

Imaan Sultan 🇵🇰🇸🇦@imaanxsultan

The Agent Economy Gala just maxed out at 400 guests  

Here’s a breakdown: 
- 203 founders ($1B+ raised combined) 
- 67 founding operators 
- 58 investors 
- 23 creators  
- 295 guests building agent applications or infrastructure  

Founders @llama_index, @daytonaio, @Tiny_Fish, @obvious, @thelinqapp  

Leads at @FireworksAI_HQ, @mercor, @mintlify, @WisprFlow  

High levels at OpenAI, Google, Meta, NVIDIA, Microsoft, Netflix, Tesla, and SpaceXAI  

But honestly none of that matters. What matters is there's no fluff. Prioritizing a quality experience, not turnout, because turnout doesn't matter when it's not relevant to the space   

EVERY person in this room is a main player, actively building out the agent ecosystem

If you still want in on the biggest event of TECH WEEK by a16z, reach out to anyone at Lemma, Linq, LlamaIndex, TinyFish, or Obvious  

See you tomorrow 🥂

🔗 View Quoted Tweet

💬2🔄1❤️6👀3731📊3

⚡ Powered by xgo.ing

fleet week makes me patriotic af

Jerry Liu (@jerryjliu0) — Fri, 09 Oct 2026 00:22:58 GMT

fleet week makes me patriotic af

Tim Baer@TimBaer8

Love WFH during Fleet Week in the nontechnical north. Finally a break from sirens, dogs, and handymen. Just weapons of war flying low over my apartment while I try to shout over them to sell AI models for doc parsing on Zoom.

🔗 View Quoted Tweet

💬1🔄0❤️17👀3008📊3

⚡ Powered by xgo.ing

LightOn OCR-3 is now available on OpenDocRouter, priced at $0.28 / 1m input and $1.40 / 1m output (i...

Jerry Liu (@jerryjliu0) — Thu, 08 Oct 2026 23:52:40 GMT

LightOn OCR-3 is now available on OpenDocRouter, priced at $0.28 / 1m input and $1.40 / 1m output (it is ~$3.19 / 1k pages on ParseBench). 

We benchmarked it on ParseBench. It is on the pareto frontier for open-weight OCR models, with similar performance as Gemini 3.8 flash low at ~45% lower price. It is decent at tables, workable for charts, and quite good at grounding.

Try it out on OpenDocRouter: opendocrouter.ai/models

For the full list of benchmark results, check out ParseBench: parsebench.ai

LightOn@LightOnIO

World-leading document recognition. Beyond text extraction. 

Introducing LightOnOCR-3, our new family of state-of-the-art AI models for document intelligence. 
Building on LightOnOCR-2, it recognizes text and handwriting, describes images, extracts data from charts and preserves document structure, all in a single processing step. 

🏆 Leading benchmark results : ahead of Chandra-OCR-2, Mistral OCR 4.1 and dots.mocr on OlmOCR-Bench, and first among open-weight models on ParseBench.

 📊 More information from every document : turn text, tables, images and charts into content your business applications can use. 

⚡Compact and efficient: 0.8B and 4B versions, with document processing up to twice as fast as LightOnOCR-2. 

🔓Apache 2.0 license: use, adapt and deploy the models on your own infrastructure. 

From handwritten administrative files to financial reports and technical documents, make more of your information searchable and usable. 

👉 Explore the models and benchmark resulhuggingface.co/blog/lightonai…o9XA

👉🏻Explore the model page on our Webslighton.ai/research/light…Mp#LightOni#DocumentIntelligencei#OCRe #OCR

🔗 View Quoted Tweet

💬8🔄9❤️94👀7067📊26

⚡ Powered by xgo.ing

Markdown has become a universal representation of information between humans and agents. It preserv...

Jerry Liu (@jerryjliu0) — Thu, 08 Oct 2026 17:52:04 GMT

Markdown has become a universal representation of information between humans and agents.

It preserves core primitives needed to express structure (headings, lists), allows agents to read the information in a semantically interpretable manner, and allows for complex elements like tables and charts. You can natively embed HTML as well.

But since most unstructured data is not natively within markdown (yet), the issue becomes the translation layer.

We've created best models to convert any unstructured document container into markdown.

Check out our blog!  llamaindex.ai/blog/markdown-…

LlamaIndex 🦙@llama_index

Markdown is all you need. 

(Mostly.)

A parser can get every word on the page right and still lose which column a number belongs to. Then your model has to guess.

Markdown keeps headings, lists, and tables intact, and it stays readable when you're debugging a bad answer. For tables with merged headers, we switch to HTML.

Read our breakdown on why it's our default output for parsing below! ⬇️

🔗 View Quoted Tweet

💬18🔄2❤️51👀4031📊24

⚡ Powered by xgo.ing

OpenDocRouter is basically OpenRouter for document OCR You can compare and use all your favorite O...

Jerry Liu (@jerryjliu0) — Thu, 08 Oct 2026 17:46:10 GMT

OpenDocRouter is basically OpenRouter for document OCR 

You can compare and use all your favorite OCR models under a single API / billing interface. 

They range from some of the lightweight/OSS models  (e.g. MinerU) to the latest frontier VLMs (Opus 5.5) 

Come check it out! opendocrouter.ai

Your browser does not support the video tag.

🔗 View on Twitter

Jerry Liu@jerryjliu0

Today I’m excited to introduce OpenDocRouter - a unified API for document parsing with the latest frontier and open-weight models.

There are a lot of VLMs and OCR models that can be used for document parsing: we have over 130+ models on ParseBench, and a HuggingFace search for “ocr” turns up thousands of results.

It’s extremely time consuming to choose between OCR vendors. You need to figure out the right prompts, handle rate limits, manage deployments, integrations with all models you’re using, and benchmark new models as they come out.

OpenDocRouter provides a comprehensive, transparent set of models along the price-performance frontier. It manages a unified API to transcribe documents to markdown. It serves all frontier and open-weight models “at-cost”, with a small transaction cut. It handles rate limits with all models to ensure you can put massive volume through. It even offers bounding boxes and layout as a service, so that you can add grounding to any model that you’re using.

When new OCR candidate models ship, we will benchmark them on ParseBench and immediately add them to OpenDocRouter. 

We’re adding a lot more models very quickly, and also adding some extremely exciting feature improvements (e.g. automated routing) as we speak.

We welcome your feedback!

Check it out: opendocrouter.ai

Blog: llamaindex.ai/blog/introduci…

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬14🔄6❤️79👀5868📊30

⚡ Powered by xgo.ing

this was a really good business decision and ensures i'll continue using @bot

Jerry Liu (@jerryjliu0) — Thu, 08 Oct 2026 05:44:52 GMT

this was a really good business decision and ensures i'll continue using @bot

Elon Musk@elonmusk

Important note regarding Grok @Bot:

Going forward, @SpaceX will use the best back end model for any given task, including Claude Opus 5.5, MidJourney, Suno and other leading APIs. 

Whatever is most likely to give you the best outcome.

🔗 View Quoted Tweet

💬8🔄1❤️12👀2880📊8

⚡ Powered by xgo.ing

having high agency is choosing to do a hot ones challenge with no spice tolerance and powering throu...

Jerry Liu (@jerryjliu0) — Thu, 08 Oct 2026 05:19:50 GMT

having high agency is choosing to do a hot ones challenge with no spice tolerance and powering through 9 out of 10 wings 

s/o matisse

💬4🔄1❤️11👀2765📊6

⚡ Powered by xgo.ing

in the meantime if you need consistently high document parsing quality across every dimension, come ...

Jerry Liu (@jerryjliu0) — Wed, 07 Oct 2026 21:04:19 GMT

in the meantime if you need consistently high document parsing quality across every dimension, come check out llamaparse: cloud.llamaindex.ai

💬0🔄0❤️1👀738📊1

⚡ Powered by xgo.ing

Claude Haiku 5.5 is now available on OpenDocRouter as one of the cheapest models for document parsin...

Jerry Liu (@jerryjliu0) — Wed, 07 Oct 2026 21:04:18 GMT

Claude Haiku 5.5 is now available on OpenDocRouter as one of the cheapest models for document parsing.

It is roughly $1.2 / 1k pages and does quite well on tables and text reading order compared to other models at this price point. It struggles a bit on charts, semantic formatting, and bounding boxes.

Check out OpenDocRouter, which we just launched today! opendocrouter.ai/models

Claude@claudeai

Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released.

On average, it costs around 75% less to run than Claude Haiku 4.5.

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬22🔄9❤️57👀5549📊27

⚡ Powered by xgo.ing

Today I’m excited to introduce OpenDocRouter - a unified API for document parsing with the latest fr...

Jerry Liu (@jerryjliu0) — Wed, 07 Oct 2026 16:04:36 GMT

Today I’m excited to introduce OpenDocRouter - a unified API for document parsing with the latest frontier and open-weight models.

There are a lot of VLMs and OCR models that can be used for document parsing: we have over 130+ models on ParseBench, and a HuggingFace search for “ocr” turns up thousands of results.

It’s extremely time consuming to choose between OCR vendors. You need to figure out the right prompts, handle rate limits, manage deployments, integrations with all models you’re using, and benchmark new models as they come out.

OpenDocRouter provides a comprehensive, transparent set of models along the price-performance frontier. It manages a unified API to transcribe documents to markdown. It serves all frontier and open-weight models “at-cost”, with a small transaction cut. It handles rate limits with all models to ensure you can put massive volume through. It even offers bounding boxes and layout as a service, so that you can add grounding to any model that you’re using.

When new OCR candidate models ship, we will benchmark them on ParseBench and immediately add them to OpenDocRouter. 

We’re adding a lot more models very quickly, and also adding some extremely exciting feature improvements (e.g. automated routing) as we speak.

We welcome your feedback!

Check it out: opendocrouter.ai

Blog: llamaindex.ai/blog/introduci…

Your browser does not support the video tag.

🔗 View on Twitter

LlamaIndex 🦙@llama_index

Today we're announcing OpenDocRouter: every model for document parsing under one API.

There are ~4,000 OCR models on Hugging Face, and the frontier labs ship a new one nearly every month. Whatever's best for your docs today won't be by Q1.

So stop asking which model to use for doc parsing. Ask how fast you can switch.

✅ Switch models in one line: same request, same markdown output, no new prompts or integrations
✅ Any model you want: 10 frontier and open-source models at launch, including Claude Opus 5.5, Gemini 3.8 Flash, GPT-6 Luna, MinerU2.5-Pro and PaddleOCR-VL-1.6
✅ Choose with receipts: every model scored on ParseBench for quality and cost
✅ Traceable output from any model: "layout: true" adds grounded bounding boxes and the same layout classes, even for models that don't support it natively
✅ Pay only for what works: per-token pricing, failed pages never charged, top up from $25

The spread is the point: $0.86 to $48.82 per 1,000 pages, depending on what your documents actually need.

Live now → opendocrouter.ai

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬62🔄51❤️514👀49862📊192

⚡ Powered by xgo.ing

You may ask why we’re doing this given that our main focus within LlamaParse (https://t.co/XYZmx5TFz...

Jerry Liu (@jerryjliu0) — Wed, 07 Oct 2026 16:04:36 GMT

You may ask why we’re doing this given that our main focus within LlamaParse (cloud.llamaindex.ai) is building the best frontier models for document parsing and extraction.

While we believe our models are best in class in accuracy at their price points, we acknowledge that they don’t cover every possible price point; at the low end there are extremely cheap models (mineru, luna, paddleocr), at the high end there may be an edge case or two you want to run the latest frontier model over.

💬0🔄0❤️9👀1039📊1

⚡ Powered by xgo.ing

You can literally do anything these days if you just ask for it

Jerry Liu (@jerryjliu0) — Wed, 07 Oct 2026 03:47:24 GMT

You can literally do anything these days if you just ask for it

💬27🔄10❤️126👀23121📊48

⚡ Powered by xgo.ing

OCR is the perfect example of a use case that has been dominated by legacy/brittle systems, and can ...

Jerry Liu (@jerryjliu0) — Wed, 07 Oct 2026 01:16:44 GMT

OCR is the perfect example of a use case that has been dominated by legacy/brittle systems, and can be solved both accurately and cheaply by applying the right amount of agentic intelligence.

A properly tuned agentic OCR needs to dynamically apply extra compute to complex elements, review and correct failures, and create the right semantic meaning throughout the page. It's a task that frontier models are both overengineered for in terms of cost and latency, and struggle at over a long tail of complex edge cases. 

Check out this blog post by @LoganMarkewich below: llamaindex.ai/blog/ocr-is-de…

LlamaIndex 🦙@llama_index

OCR is dead 🪦 Long live agentic OCR!

Traditional OCR makes one pass and hands back whatever text it got. Tables get flattened, charts disappear, and multi-column layouts come out scrambled. Nothing checks the output.

Agentic OCR treats parsing as a loop instead:

✅ layout-aware reading order
✅ smart routing of hard elements to the right model
✅ multi-pass verification and self-correction
✅ multimodal parsing of charts, images, and complex tables

Logan Markewich, LlamaIndex's Head of Open Source, wrote about what that shift looks like in practice and where it's still struggling. Link to the breakdown in the comments below.

🔗 View Quoted Tweet

💬26🔄13❤️174👀13122📊67

⚡ Powered by xgo.ing

I agree that chatgpt/codex has the current best agent interface for deep work. - It unifies all my ...

Jerry Liu (@jerryjliu0) — Sun, 04 Oct 2026 20:29:11 GMT

I agree that chatgpt/codex has the current best agent interface for deep work. 
- It unifies all my work (coding/knowledge work) in one interface. Toggling between chatgpt/codex doesn't have explicitly separate flows
- It has forking (why doesn't the claude app have forking?) 

Claude Code CLI is probably the best that a CLI app can get. I still love Opus 5.5, especially for product demos, but mostly use it through the CLI. But sometimes having a GUI is nicer.

Yuchen Jin@Yuchenj_UW

I haven’t touched Claude Code or Codex CLI in a while.

The terminal era is over imo. It's the wrong interface for coding agents. Tabs are ephemeral, but context is persistent, and managing 30 tabs is pure cognitive overhead.

I don’t really need an IDE like Cursor either. I rarely navigate the whole codebase anymore.

The new primitive is the agent, not the file.

(Codex desktop app is the best agentic UI for now. But we’re still early.)

🔗 View Quoted Tweet

💬26🔄0❤️44👀7860📊27

⚡ Powered by xgo.ing

We got extremely good at extracting a scanned from layer by layer 📋 - from printed fields, overlaid...

Jerry Liu (@jerryjliu0) — Sun, 04 Oct 2026 19:53:39 GMT

We got extremely good at extracting a scanned from layer by layer 📋 - from printed fields, overlaid text, handwriting, and reviewer annotations.

For instance, on an annotated permit, our Extract agents can extract out both the original permit number but also any drawn/handwritten annotations on top. they're also grounded exactly in the source text. 

our agents reduce error rates by 2x+, at a small fraction of the price compared to using a frontier model.

Our Extract v2.5 agents are SOTA in price/performance, and scanned forms are just one example. Come check it out! 

Extract v2.5:llamaindex.ai/blog/introduci…3

Available in LlamaParse:cloud.llamaindex.ai8

Your browser does not support the video tag.

🔗 View on Twitter

Jerry Liu@jerryjliu0

Today we’re introducing Extract v2.5 - a series of frontier agents tuned for document extraction.

The agents (cost-effective, agentic, agentic plus) are tuned for value accuracy and grounding. Our extraction agents outperform Opus 5.5 and GPT-6 Sol while being 30%-4x cheaper. 

We’ve made massive improvements on complex extraction over 
* long lists (86.1% -> 95.5% on our agentic tier)
* records spanning pages (85.5% -> 96.5%)
* scanned forms (90.9% -> 95.7%)

We’ve also launched the following features:
✅ Advanced citations: we locate bounding boxes for all supporting values for an inferred field, even if there's not an exact match.
✅ Structural Reasoning: we tailor document extraction algorithms depending on the type, layout, and information

Our extraction agents are SOTA in price-performance on ExtractBench, across a wide range of cost points. We are the best tool for document extraction across documents of any complexity.

Blog: llamaindex.ai/blog/introduci…

All of these are available on LlamaParse: cloud.llamaindex.ai . Come check it out!

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬24🔄13❤️119👀11860📊49

⚡ Powered by xgo.ing

sfo is the best airport in the US

Jerry Liu (@jerryjliu0) — Sun, 04 Oct 2026 03:10:23 GMT

sfo is the best airport in the US

justin@justinsunyt

sfo probably the only airport in the world where you can leave your house an hour and a half before your international flight and still have time to spare before boarding

🔗 View Quoted Tweet

💬22🔄4❤️235👀26350📊40

⚡ Powered by xgo.ing

Our new extraction agents are extremely good at reasoning over tables that span multiple pages some...

Jerry Liu (@jerryjliu0) — Sat, 03 Oct 2026 01:01:11 GMT

Our new extraction agents are extremely good at reasoning over tables that span multiple pages

sometimes a record starts on one page and finishes on the next. extremely common for real world docs (insurance claims, regulatory filings, legal schedules) - for instance you print Word / Google Doc to PDF 

our agents can piece these cells along with hundreds/thousands of other cells into clean, structured tabular output.

come check it out: llamaindex.ai/blog/introduci…

come sign up to llamaparse: cloud.llamaindex.ai

Your browser does not support the video tag.

🔗 View on Twitter

Jerry Liu@jerryjliu0

Today we’re introducing Extract v2.5 - a series of frontier agents tuned for document extraction.

The agents (cost-effective, agentic, agentic plus) are tuned for value accuracy and grounding. Our extraction agents outperform Opus 5.5 and GPT-6 Sol while being 30%-4x cheaper. 

We’ve made massive improvements on complex extraction over 
* long lists (86.1% -> 95.5% on our agentic tier)
* records spanning pages (85.5% -> 96.5%)
* scanned forms (90.9% -> 95.7%)

We’ve also launched the following features:
✅ Advanced citations: we locate bounding boxes for all supporting values for an inferred field, even if there's not an exact match.
✅ Structural Reasoning: we tailor document extraction algorithms depending on the type, layout, and information

Our extraction agents are SOTA in price-performance on ExtractBench, across a wide range of cost points. We are the best tool for document extraction across documents of any complexity.

Blog: llamaindex.ai/blog/introduci…

All of these are available on LlamaParse: cloud.llamaindex.ai . Come check it out!

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬14🔄5❤️49👀6100📊20

⚡ Powered by xgo.ing

one of the most challenging tasks for frontier models is being able to extract thousands of values f...

Jerry Liu (@jerryjliu0) — Fri, 02 Oct 2026 15:29:07 GMT

one of the most challenging tasks for frontier models is being able to extract thousands of values from extremely dense tables in documents.

our new Extract v2.5 agents are able to get 93%-96%+ on long-list extraction, including cells that fall in between pages. 

in contrast, astra gets ~30% accuracy over this benchmark. 

in these cases, frontier VLMs can stop early or miss repeated records. 

they also struggle to attribute each value back to the source. we are able to attribute every single extracted value back to the source.

check out the video below, the blog post, and llamaparse! 

blog: llamaindex.ai/blog/introduci…

llamaparse: cloud.llamaindex.ai

Your browser does not support the video tag.

🔗 View on Twitter

Jerry Liu@jerryjliu0

Today we’re introducing Extract v2.5 - a series of frontier agents tuned for document extraction.

The agents (cost-effective, agentic, agentic plus) are tuned for value accuracy and grounding. Our extraction agents outperform Opus 5.5 and GPT-6 Sol while being 30%-4x cheaper. 

We’ve made massive improvements on complex extraction over 
* long lists (86.1% -> 95.5% on our agentic tier)
* records spanning pages (85.5% -> 96.5%)
* scanned forms (90.9% -> 95.7%)

We’ve also launched the following features:
✅ Advanced citations: we locate bounding boxes for all supporting values for an inferred field, even if there's not an exact match.
✅ Structural Reasoning: we tailor document extraction algorithms depending on the type, layout, and information

Our extraction agents are SOTA in price-performance on ExtractBench, across a wide range of cost points. We are the best tool for document extraction across documents of any complexity.

Blog: llamaindex.ai/blog/introduci…

All of these are available on LlamaParse: cloud.llamaindex.ai . Come check it out!

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬13🔄10❤️116👀13464📊41

⚡ Powered by xgo.ing

Today we’re introducing Extract v2.5 - a series of frontier agents tuned for document extraction. T...

Jerry Liu (@jerryjliu0) — Thu, 01 Oct 2026 16:12:51 GMT

Today we’re introducing Extract v2.5 - a series of frontier agents tuned for document extraction.

The agents (cost-effective, agentic, agentic plus) are tuned for value accuracy and grounding. Our extraction agents outperform Opus 5.5 and GPT-6 Sol while being 30%-4x cheaper. 

We’ve made massive improvements on complex extraction over 
* long lists (86.1% -> 95.5% on our agentic tier)
* records spanning pages (85.5% -> 96.5%)
* scanned forms (90.9% -> 95.7%)

We’ve also launched the following features:
✅ Advanced citations: we locate bounding boxes for all supporting values for an inferred field, even if there's not an exact match.
✅ Structural Reasoning: we tailor document extraction algorithms depending on the type, layout, and information

Our extraction agents are SOTA in price-performance on ExtractBench, across a wide range of cost points. We are the best tool for document extraction across documents of any complexity.

Blog: llamaindex.ai/blog/introduci…

All of these are available on LlamaParse: cloud.llamaindex.ai . Come check it out!

Your browser does not support the video tag.

🔗 View on Twitter

💬42🔄35❤️291👀52360📊141