Perplexity(@perplexity_ai)

Gemini 4 Argon sets a new record on WANDR. It is now available in Perplexity for Pro and Max subscr...

Perplexity (@perplexity_ai) — Fri, 16 Oct 2026 15:03:25 GMT

Gemini 4 Argon sets a new record on WANDR.

It is now available in Perplexity for Pro and Max subscribers.

💬70🔄47❤️695👀117710📊143

⚡ Powered by xgo.ing

NVIDIA’s Nemotron 3.5 Super VL is now available in Perplexity for Pro and Max subscribers, and in Ag...

Perplexity (@perplexity_ai) — Thu, 15 Oct 2026 18:03:15 GMT

NVIDIA’s Nemotron 3.5 Super VL is now available in Perplexity for Pro and Max subscribers, and in Agent API.

Your browser does not support the video tag.

🔗 View on Twitter

💬21🔄17❤️303👀68601📊53

⚡ Powered by xgo.ing

SPACE is the first component of Perplexity's Agentic Runtime Platform. We're open sourcing it to rai...

Perplexity (@perplexity_ai) — Thu, 15 Oct 2026 17:03:11 GMT

SPACE is the first component of Perplexity's Agentic Runtime Platform. We're open sourcing it to raise the security bar for agentic products.

Read the blog: perplexity.ai/hub/blog/open-…

💬1🔄3❤️47👀8527📊8

⚡ Powered by xgo.ing

We tested SPACE by attacking it with frontier models. In Escaping SPACE: Part I, nine model configur...

Perplexity (@perplexity_ai) — Thu, 15 Oct 2026 17:03:08 GMT

We tested SPACE by attacking it with frontier models. In Escaping SPACE: Part I, nine model configurations with root inside the VM made 108 escape attempts, and none reached the host.

Read the research: perplexity.ai/hub/blog/escap…

💬1🔄0❤️43👀10478📊8

⚡ Powered by xgo.ing

We're open sourcing SPACE, the sandbox platform that runs every task in Perplexity Computer. SPACE ...

Perplexity (@perplexity_ai) — Thu, 15 Oct 2026 17:03:00 GMT

We're open sourcing SPACE, the sandbox platform that runs every task in Perplexity Computer.

SPACE gives each agent task its own Firecracker microVM and controls credentials and outbound traffic from outside the VM.

Available now under Apache 2.0: github.com/perplexityai/S…

💬146🔄215❤️2512👀306710📊737

⚡ Powered by xgo.ing

Ultrafast versions of GPT-6 Astra and SOL 6.1 are now available in Perplexity Computer for Max subsc...

Perplexity (@perplexity_ai) — Wed, 14 Oct 2026 17:01:29 GMT

Ultrafast versions of GPT-6 Astra and SOL 6.1 are now available in Perplexity Computer for Max subscribers.

Both generate tokens faster, completing long Computer tasks sooner.

Choose Astra or SOL 6.1 under Custom, then set speed to Ultrafast.

Your browser does not support the video tag.

🔗 View on Twitter

💬53🔄28❤️512👀98521📊110

⚡ Powered by xgo.ing

Terminals and file watches can outlive their opening requests. Once a receiver installs the new exec...

Perplexity (@perplexity_ai) — Wed, 14 Oct 2026 16:02:44 GMT

Terminals and file watches can outlive their opening requests. Once a receiver installs the new execution identity, SPACE stops the old streams. Eligible operations can reconnect once, passing the identity checks again.

Read the article: perplexity.ai/hub/blog/how-s…

💬1🔄0❤️16👀4707📊3

⚡ Powered by xgo.ing

Each request carries short-lived, signed proof of which execution may handle it and what access is a...

Perplexity (@perplexity_ai) — Wed, 14 Oct 2026 16:02:40 GMT

Each request carries short-lived, signed proof of which execution may handle it and what access is allowed.

The receiving worker checks that identity against its own runtime record and rejects mismatches before forwarding.

💬1🔄1❤️21👀5420📊3

⚡ Powered by xgo.ing

Our sandbox, SPACE, handles millions of executions a day with an error rate below 0.0002%. We’re sh...

Perplexity (@perplexity_ai) — Wed, 14 Oct 2026 16:02:34 GMT

Our sandbox, SPACE, handles millions of executions a day with an error rate below 0.0002%.

We’re sharing how it uses cryptographic identity to keep requests tied to the right execution as sandboxes pause, resume, and move.

Read more: perplexity.ai/hub/blog/how-s…

💬40🔄20❤️247👀57164📊79

⚡ Powered by xgo.ing

Because each sandbox holds a real checkout, agents search it with grep and standard file tools inste...

Perplexity (@perplexity_ai) — Mon, 12 Oct 2026 18:05:31 GMT

Because each sandbox holds a real checkout, agents search it with grep and standard file tools instead of relying on broad semantic searches.

On an internal benchmark of 150 financial-document questions, this reached 0.627 answer accuracy versus about 0.37 for semantic search.

💬1🔄0❤️13👀5962📊3

⚡ Powered by xgo.ing

A single long-lived sandbox per project would need provisioning for peak load and months of uptime. ...

Perplexity (@perplexity_ai) — Mon, 12 Oct 2026 18:05:27 GMT

A single long-lived sandbox per project would need provisioning for peak load and months of uptime.

Instead, repositories live on S3-backed storage and each agent runs in its own ephemeral sandbox, so a crash loses only unsubmitted work.

💬1🔄1❤️9👀6271📊2

⚡ Powered by xgo.ing

Git provides atomic multi-file commits, authorship, and queryable history. A forge hosts the reposi...

Perplexity (@perplexity_ai) — Mon, 12 Oct 2026 18:05:19 GMT

Git provides atomic multi-file commits, authorship, and queryable history.

A forge hosts the repository, controls who can update it, and adds PR review. Projects keep the first two and replace review with a check of each submit against a repository contract.

💬0🔄0❤️0

⚡ Powered by xgo.ing

Agents collaborating on a long-running project face the same problem as engineers sharing a codebase...

Perplexity (@perplexity_ai) — Mon, 12 Oct 2026 18:05:16 GMT

Agents collaborating on a long-running project face the same problem as engineers sharing a codebase.

Work must outlive ephemeral sandboxes, concurrent writes must not overwrite each other, and history must show who changed what.

💬1🔄0❤️0📊1

⚡ Powered by xgo.ing

This is how we built the filesystem behind Perplexity Projects as a Git forge for agents. Each agen...

Perplexity (@perplexity_ai) — Mon, 12 Oct 2026 18:05:13 GMT

This is how we built the filesystem behind Perplexity Projects as a Git forge for agents.

Each agent works in its own sandbox and commits to a shared repository where people and other agents can build on its work across sessions.

Read more: perplexity.ai/hub/blog/build…

💬71🔄52❤️424👀86529📊155

⚡ Powered by xgo.ing

Both models are on Hugging Face: https://t.co/JHy1DhFaCT

Perplexity (@perplexity_ai) — Wed, 07 Oct 2026 16:10:36 GMT

Both models are on Hugging Face: huggingface.co/collections/pe…

💬0🔄0❤️20👀5831📊3

⚡ Powered by xgo.ing

On MADQA, 500 questions over 800 PDFs, it reaches 92.4%, the top result among retrievers.

Perplexity (@perplexity_ai) — Wed, 07 Oct 2026 16:10:32 GMT

On MADQA, 500 questions over 800 PDFs, it reaches 92.4%, the top result among retrievers.

💬1🔄0❤️15👀5957📊3

⚡ Powered by xgo.ing

In agentic search, a GPT-OSS-120B agent using the 9B model answers 64.0% of BrowseComp+ questions co...

Perplexity (@perplexity_ai) — Wed, 07 Oct 2026 16:10:27 GMT

In agentic search, a GPT-OSS-120B agent using the 9B model answers 64.0% of BrowseComp+ questions correctly, 4.9 points ahead of the next ColBERT model, with fewer searches than any baseline.

💬1🔄1❤️13👀2366📊3

⚡ Powered by xgo.ing

On Q2D-Web, about 70,000 production-derived search queries over 190M web documents, the 9B and 0.6B ...

Perplexity (@perplexity_ai) — Wed, 07 Oct 2026 16:10:21 GMT

On Q2D-Web, about 70,000 production-derived search queries over 190M web documents, the 9B and 0.6B models reach 74.8% and 73.6% Recall@1000, against 69.3% for nemotron-embed-8b.

💬4🔄0❤️16👀2400📊5

⚡ Powered by xgo.ing

We're releasing pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images...

Perplexity (@perplexity_ai) — Wed, 07 Oct 2026 16:09:58 GMT

We're releasing pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and pages with a shared embedding space for cross-model querying.

Both models achieve frontier performance and are publicly available on Hugging Face.

perplexity.ai/hub/blog/multi…

💬72🔄121❤️683👀122736📊214

⚡ Powered by xgo.ing

Perplexity Computer now creates interactive charts and visualizations directly in your thread. For ...

Perplexity (@perplexity_ai) — Thu, 01 Oct 2026 16:12:01 GMT

Perplexity Computer now creates interactive charts and visualizations directly in your thread.

For financial data, Computer uses @tradingview Lightweight Charts for candlesticks, volume, and moving averages.

Your browser does not support the video tag.

🔗 View on Twitter

💬77🔄41❤️415👀77460📊123

⚡ Powered by xgo.ing

View pplx-embed-v2-context-9b-preview on Hugging Face: https://t.co/EeqI2xCFOl

Perplexity (@perplexity_ai) — Wed, 30 Sep 2026 19:07:51 GMT

View pplx-embed-v2-context-9b-preview on Hugging Face:
huggingface.co/perplexity-ai/…

💬1🔄2❤️33👀7021📊9

⚡ Powered by xgo.ing

On ConTEB, the preview has the highest average nDCG@10 of the models tested, though not on every tas...

Perplexity (@perplexity_ai) — Wed, 30 Sep 2026 19:07:48 GMT

On ConTEB, the preview has the highest average nDCG@10 of the models tested, though not on every task.

It also beats voyage-context-4 on chunk retrieval while using 8x less storage per vector: 1 KB (1024 dims, int8) vs 8 KB (2048, float32).

💬5🔄2❤️17👀7590📊6

⚡ Powered by xgo.ing

We built a new way to train contextual embedding models, which encode each chunk of a document with ...

Perplexity (@perplexity_ai) — Wed, 30 Sep 2026 19:07:29 GMT

We built a new way to train contextual embedding models, which encode each chunk of a document with the whole document in view.

pplx-embed-v2-context-9b-preview sets a new state of the art on ConTEB and @turbopuffer's new, privately held context-bench.

perplexity.ai/hub/blog/conte…

💬58🔄66❤️718👀259403📊208

⚡ Powered by xgo.ing

Introducing Automations in Perplexity Computer. Automations are for ongoing work. They can take act...

Perplexity (@perplexity_ai) — Tue, 29 Sep 2026 16:46:10 GMT

Introducing Automations in Perplexity Computer.

Automations are for ongoing work. They can take action in response to event-based triggers or on a schedule.

Automations work with your memory, skills, and connected apps like Slack, Gmail, Outlook, and Linear.

Your browser does not support the video tag.

🔗 View on Twitter

💬54🔄74❤️787👀156970📊188

⚡ Powered by xgo.ing

Our Secure Intelligence Institute works with researchers at Stanford, CMU, Duke, Columbia, Ohio Stat...

Perplexity (@perplexity_ai) — Tue, 29 Sep 2026 15:12:57 GMT

Our Secure Intelligence Institute works with researchers at Stanford, CMU, Duke, Columbia, Ohio State, and UVA.

We’re also working with NVIDIA and the Open Secure AI Alliance, and sharing tools and findings so others can strengthen their own systems.

blogs.nvidia.com/blog/open-secu…

💬0🔄2❤️10👀4714📊2

⚡ Powered by xgo.ing

Bumblebee checks developer machines for risky packages and extensions. Computer reviews findings fr...

Perplexity (@perplexity_ai) — Tue, 29 Sep 2026 15:12:54 GMT

Bumblebee checks developer machines for risky packages and extensions.

Computer reviews findings from Bumblebee and Numbat and proposes better detection rules. Humans approve every change before it ships.

Agents help improve the controls but can't approve their own changes.

Perplexity@perplexity_ai

Today we're open-sourcing Bumblebee, a read-only scanner for macOS and Linux.

It checks developer machines for risky packages, extensions, and AI tool configs.

Connected to Computer, it can trigger deeper scans whenever a new supply-chain risk emerges.

github.com/perplexityai/b…

🔗 View Quoted Tweet

💬2🔄1❤️12👀7300📊3

⚡ Powered by xgo.ing

Agent governance is an engineering problem. We’ve built safeguards into Perplexity’s infrastructure...

Perplexity (@perplexity_ai) — Tue, 29 Sep 2026 15:12:37 GMT

Agent governance is an engineering problem.

We’ve built safeguards into Perplexity’s infrastructure, harnesses, and tools, and put them to work across our products.

Today we're sharing how we engineer safer agents:
perplexity.ai/hub/blog/how-w…

💬32🔄42❤️300👀95815📊84

⚡ Powered by xgo.ing

We evaluated 10 third-party sandbox platforms and found that 8 were vulnerable to similar network-po...

Perplexity (@perplexity_ai) — Mon, 28 Sep 2026 15:10:39 GMT

We evaluated 10 third-party sandbox platforms and found that 8 were vulnerable to similar network-policy bypasses.

We disclosed all findings to the affected vendors and heard back from each.

💬3🔄3❤️24👀5241📊7

⚡ Powered by xgo.ing

Claude Opus 5.0 identified the IP-sharing route and declined to use it, reading our instruction not ...

Perplexity (@perplexity_ai) — Mon, 28 Sep 2026 15:10:35 GMT

Claude Opus 5.0 identified the IP-sharing route and declined to use it, reading our instruction not to target other external systems as ruling out third-party services.

Fable and GPT-6 Astra refused the tasks outright.

💬2🔄2❤️22👀5163📊5

⚡ Powered by xgo.ing

We’re partnering with Nvidia and 100+ industry partners to build infrastructure that contains rogue ...

Perplexity (@perplexity_ai) — Mon, 28 Sep 2026 15:10:12 GMT

We’re partnering with Nvidia and 100+ industry partners to build infrastructure that contains rogue AI agents.

In this research, we gave 9 AI models root access inside SPACE and told them to break out.

Across 108 runs, none breached the VM boundary.
perplexity.ai/hub/blog/escap…

💬61🔄56❤️433👀150077📊118

⚡ Powered by xgo.ing

Inside Photon, p99 response time dropped from about 800 ms to about 65 ms. It uses about 20% fewer s...

Perplexity (@perplexity_ai) — Thu, 24 Sep 2026 18:08:12 GMT

Inside Photon, p99 response time dropped from about 800 ms to about 65 ms. It uses about 20% fewer serving machines and stores 2.5x as much data per document.

How we built Photon: perplexity.ai/hub/blog/photon
Get started with Fast Search: pplx.ai/fast-search-api

💬3🔄0❤️26👀7582📊7

⚡ Powered by xgo.ing

Photon builds new indexes separately from the servers handling live queries. When a version is read...

Perplexity (@perplexity_ai) — Thu, 24 Sep 2026 18:08:09 GMT

Photon builds new indexes separately from the servers handling live queries.

When a version is ready, Photon updates one serving group at a time, warms its caches with real search queries, checks readiness, and returns it to routing. The other groups keep serving traffic.

💬3🔄0❤️16👀8092📊4

⚡ Powered by xgo.ing

Introducing Fast Search in the Perplexity Search API. Fast Search runs on Photon, our new Rust-base...

Perplexity (@perplexity_ai) — Thu, 24 Sep 2026 18:07:51 GMT

Introducing Fast Search in the Perplexity Search API.

Fast Search runs on Photon, our new Rust-based retrieval and ranking service that we built with a small team of engineers and hundreds of agents.

It returns 95% of search results in 230 ms or less.

💬66🔄39❤️551👀149318📊138

⚡ Powered by xgo.ing

Portable Computer runs in the Perplexity app for Windows and is available to all Consumer and Enterp...

Perplexity (@perplexity_ai) — Thu, 24 Sep 2026 16:35:41 GMT

Portable Computer runs in the Perplexity app for Windows and is available to all Consumer and Enterprise Pro and Max subscribers.

Get started: perplexity.ai/hub/products/p…

💬0🔄3❤️15👀7779📊3

⚡ Powered by xgo.ing

Portable Computer works across connected apps like Slack, Gmail, and Google Drive, along with files ...

Perplexity (@perplexity_ai) — Thu, 24 Sep 2026 16:35:38 GMT

Portable Computer works across connected apps like Slack, Gmail, and Google Drive, along with files on your device.

Running tasks entirely locally keeps your files, data, and answers on device and doesn’t use Computer credits.

💬3🔄3❤️22👀7614📊6

⚡ Powered by xgo.ing

Portable Computer for Windows is now available on @AMD Ryzen AI Max Series processors. It makes it ...

Perplexity (@perplexity_ai) — Thu, 24 Sep 2026 16:35:30 GMT

Portable Computer for Windows is now available on @AMD Ryzen AI Max Series processors.

It makes it easy to run local AI agents that work with your connected apps and local files. Kick off tasks or schedule recurring work that runs entirely on your device.

Your browser does not support the video tag.

🔗 View on Twitter

💬47🔄52❤️378👀77266📊96

⚡ Powered by xgo.ing

Read the full article: https://t.co/bcL0B2KQyd

Perplexity (@perplexity_ai) — Tue, 22 Sep 2026 20:22:17 GMT

Read the full article:

perplexity.ai/hub/blog/learn…

💬3🔄3❤️13👀6800📊7

⚡ Powered by xgo.ing

With hints, the unchanged model avoided the original failure in 93.7% of cases, up from 75.1% In a ...

Perplexity (@perplexity_ai) — Tue, 22 Sep 2026 20:22:14 GMT

With hints, the unchanged model avoided the original failure in 93.7% of cases, up from 75.1%

In a separate live test, tool-call failures fell from 2.24% to 1.77% between trained versions, without hints at inference time.

💬4🔄2❤️12👀7261📊5

⚡ Powered by xgo.ing

New research: We post-trained a Computer model to learn from its own errors using hint-guided self-d...

Perplexity (@perplexity_ai) — Tue, 22 Sep 2026 20:21:54 GMT

New research: We post-trained a Computer model to learn from its own errors using hint-guided self-distillation.

In a live A/B test, a later trained checkpoint reduced tool-call failures by 21.2% relative to an earlier checkpoint.

💬40🔄32❤️386👀113948📊98

⚡ Powered by xgo.ing

GPT-6 Sol is now available in Perplexity and Computer. It's now the default Light option in Compute...

Perplexity (@perplexity_ai) — Tue, 22 Sep 2026 19:08:10 GMT

GPT-6 Sol is now available in Perplexity and Computer.

It's now the default Light option in Computer's effort selector.

💬25🔄23❤️386👀28115📊63

⚡ Powered by xgo.ing

Claude Opus 5.5 is now available as the Standard effort level in Perplexity Computer. On our WANDR ...

Perplexity (@perplexity_ai) — Tue, 22 Sep 2026 16:48:21 GMT

Claude Opus 5.5 is now available as the Standard effort level in Perplexity Computer.

On our WANDR benchmark, it scored 0.610 at $4.13 per task, slightly outperforming Claude Fable 5.1 while costing 67.6% less per task.

💬25🔄21❤️418👀33466📊65

⚡ Powered by xgo.ing

Perplexity Computer can now create videos using MiniMax H3 and ByteDance Seedance 2.5. Ask Compute...

Perplexity (@perplexity_ai) — Mon, 21 Sep 2026 16:25:45 GMT

Perplexity Computer can now create videos using MiniMax H3 and ByteDance Seedance 2.5. 

Ask Computer for a campaign clip, product demo, or social asset, and it produces finished video next to copy and creative in same thread. 

Available now for Pro and Max subscribers.

Your browser does not support the video tag.

🔗 View on Twitter

💬61🔄64❤️785👀141349📊170

⚡ Powered by xgo.ing

Computer’s goal is maximum intelligence at minimum cost through multi-model orchestration. We’ll co...

Perplexity (@perplexity_ai) — Thu, 17 Sep 2026 16:08:24 GMT

Computer’s goal is maximum intelligence at minimum cost through multi-model orchestration.

We’ll continue optimizing the model-agnostic harness and refining the orchestrator and subagent combinations behind each effort level as frontier models advance.

perplexity.ai/hub/blog/compu…

💬8🔄0❤️38👀7611📊10

⚡ Powered by xgo.ing

For more control, you can customize model selection, reasoning level, and available speed settings.

Perplexity (@perplexity_ai) — Thu, 17 Sep 2026 16:08:21 GMT

For more control, you can customize model selection, reasoning level, and available speed settings.

💬2🔄2❤️54👀11986📊7

⚡ Powered by xgo.ing

We’re rolling out effort controls in Computer’s model selector. Effort presets combine the orchestr...

Perplexity (@perplexity_ai) — Thu, 17 Sep 2026 16:08:17 GMT

We’re rolling out effort controls in Computer’s model selector.

Effort presets combine the orchestrator model and reasoning depth to control how deeply and efficiently Computer works through a task.

Available now on web. Coming to mobile and desktop soon.

Your browser does not support the video tag.

🔗 View on Twitter

💬71🔄47❤️533👀173469📊135

⚡ Powered by xgo.ing

CobbleDB is purpose-built for repeated batch reads of prepared page records, optimizing performance ...

Perplexity (@perplexity_ai) — Tue, 15 Sep 2026 20:17:14 GMT

CobbleDB is purpose-built for repeated batch reads of prepared page records, optimizing performance and cost for Perplexity’s search workload.

We plan to open-source CobbleDB for other teams building AI search.

Read more: perplexity.ai/hub/blog/cobbl…

💬11🔄10❤️83👀13239📊25

⚡ Powered by xgo.ing

After replacing DynamoDB, production before-and-after measurements show median batch-read latency fa...

Perplexity (@perplexity_ai) — Tue, 15 Sep 2026 20:17:11 GMT

After replacing DynamoDB, production before-and-after measurements show median batch-read latency falling from 31.4 to 5.60 ms and p99 from 123 to 24.2 ms.

Our internal cost model estimates at least 20% savings relative to DynamoDB.

💬3🔄2❤️43👀20272📊9

⚡ Powered by xgo.ing

We’re publishing research on how we built CobbleDB, our key-value database that serves web content f...

Perplexity (@perplexity_ai) — Tue, 15 Sep 2026 20:16:51 GMT

We’re publishing research on how we built CobbleDB, our key-value database that serves web content for Perplexity search.

Two engineers and a team of hundreds of proactive, always-on AI agents built the core infrastructure in two months.

💬84🔄92❤️966👀498183📊260

⚡ Powered by xgo.ing

Perplexity will come pre-loaded starting with the HP ZBook Ultra G3A and will expand to more HP devi...

Perplexity (@perplexity_ai) — Tue, 15 Sep 2026 15:02:27 GMT

Perplexity will come pre-loaded starting with the HP ZBook Ultra G3A and will expand to more HP devices over time.

Read more:
perplexity.ai/hub/blog/perpl…

💬0🔄2❤️12👀5354📊2

Perplexity(@perplexity_ai)

Gemini 4 Argon sets a new record on WANDR. It is now available in Perplexity for Pro and Max subscr...

NVIDIA’s Nemotron 3.5 Super VL is now available in Perplexity for Pro and Max subscribers, and in Ag...

SPACE is the first component of Perplexity's Agentic Runtime Platform. We're open sourcing it to rai...

We tested SPACE by attacking it with frontier models. In Escaping SPACE: Part I, nine model configur...

We're open sourcing SPACE, the sandbox platform that runs every task in Perplexity Computer. SPACE ...

Ultrafast versions of GPT-6 Astra and SOL 6.1 are now available in Perplexity Computer for Max subsc...

Terminals and file watches can outlive their opening requests. Once a receiver installs the new exec...

Each request carries short-lived, signed proof of which execution may handle it and what access is a...

Our sandbox, SPACE, handles millions of executions a day with an error rate below 0.0002%. We’re sh...

Because each sandbox holds a real checkout, agents search it with grep and standard file tools inste...

A single long-lived sandbox per project would need provisioning for peak load and months of uptime. ...

Git provides atomic multi-file commits, authorship, and queryable history. A forge hosts the reposi...

Agents collaborating on a long-running project face the same problem as engineers sharing a codebase...

This is how we built the filesystem behind Perplexity Projects as a Git forge for agents. Each agen...

Both models are on Hugging Face: https://t.co/JHy1DhFaCT

On MADQA, 500 questions over 800 PDFs, it reaches 92.4%, the top result among retrievers.

In agentic search, a GPT-OSS-120B agent using the 9B model answers 64.0% of BrowseComp+ questions co...

On Q2D-Web, about 70,000 production-derived search queries over 190M web documents, the 9B and 0.6B ...

We're releasing pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images...

Perplexity Computer now creates interactive charts and visualizations directly in your thread. For ...

View pplx-embed-v2-context-9b-preview on Hugging Face: https://t.co/EeqI2xCFOl

On ConTEB, the preview has the highest average nDCG@10 of the models tested, though not on every tas...

We built a new way to train contextual embedding models, which encode each chunk of a document with ...

Read more about Automations in Perplexity Computer: https://t.co/XuNx97oJjb

Introducing Automations in Perplexity Computer. Automations are for ongoing work. They can take act...

Our Secure Intelligence Institute works with researchers at Stanford, CMU, Duke, Columbia, Ohio Stat...

Bumblebee checks developer machines for risky packages and extensions. Computer reviews findings fr...

Agent governance is an engineering problem. We’ve built safeguards into Perplexity’s infrastructure...

We evaluated 10 third-party sandbox platforms and found that 8 were vulnerable to similar network-po...

Claude Opus 5.0 identified the IP-sharing route and declined to use it, reading our instruction not ...

We’re partnering with Nvidia and 100+ industry partners to build infrastructure that contains rogue ...

Inside Photon, p99 response time dropped from about 800 ms to about 65 ms. It uses about 20% fewer s...

Photon builds new indexes separately from the servers handling live queries. When a version is read...

Introducing Fast Search in the Perplexity Search API. Fast Search runs on Photon, our new Rust-base...

Portable Computer runs in the Perplexity app for Windows and is available to all Consumer and Enterp...

Portable Computer works across connected apps like Slack, Gmail, and Google Drive, along with files ...

Portable Computer for Windows is now available on @AMD Ryzen AI Max Series processors. It makes it ...

Read the full article: https://t.co/bcL0B2KQyd

With hints, the unchanged model avoided the original failure in 93.7% of cases, up from 75.1% In a ...

New research: We post-trained a Computer model to learn from its own errors using hint-guided self-d...

GPT-6 Sol is now available in Perplexity and Computer. It's now the default Light option in Compute...

Claude Opus 5.5 is now available as the Standard effort level in Perplexity Computer. On our WANDR ...

Perplexity Computer can now create videos using MiniMax H3 and ByteDance Seedance 2.5. Ask Compute...

Computer’s goal is maximum intelligence at minimum cost through multi-model orchestration. We’ll co...

For more control, you can customize model selection, reasoning level, and available speed settings.

We’re rolling out effort controls in Computer’s model selector. Effort presets combine the orchestr...

CobbleDB is purpose-built for repeated batch reads of prepared page records, optimizing performance ...

After replacing DynamoDB, production before-and-after measurements show median batch-read latency fa...

We’re publishing research on how we built CobbleDB, our key-value database that serves web content f...

Perplexity will come pre-loaded starting with the HP ZBook Ultra G3A and will expand to more HP devi...