Dan Shipper, CEO of Every, shares five takeaways from testing OpenAI's newest model, Astra. He walks through examples including editing a video in Premiere and building a Battle of Waterloo simulation, covering writing, computer use, 3D games and visualizations, interface design, and how well the model understands prompts, before a final verdict on when Astra earns a place in a daily workflow. # Channel Presentation Observations ## Intended Audience - **AI enthusiasts and professionals tracking model releases.** The video opens assuming day-one urgency: *"It's model release day... the long-awaited Astra is dropping today"* — implying an audience that wants immediate, same-day evaluations. - **People who want practical guidance on which AI tools to use:** *"we've got everything that you need to know which models to use and for what on day one."* ## Claimed Scope of the Channel - **Comprehensive AI coverage under one brand.** The channel is explicitly positioned as a one-stop AI resource: *"It's the only subscription you need to stay at the edge of AI. If you wanna be at the edge, you need education and you need equipment."* - **Team-based, structured testing operation.** The creator claims organizational scale behind reviews: *"We've got a 30-person team taking it through everything from writing, to coding, to design."* - **Original product development.** Beyond commentary, the channel claims it builds tools: *"we build our own tools from an email agent like Cora to the Every agent, an agentic coworker that lives in your Slack."* ## Self-Description Evidence (Creator Claims) - **Brand identity: "Every" (every.to)** — explicitly named and promoted repeatedly, bundled as *"one subscription"* including *"a daily newsletter... courses... camps."* - **Depth promise:** *"we have the full vibe check. It's a several thousand word article with all of our testing from our whole team live right now on the Every website."* - **Distinctive format: "vibe checks."** This video and prior releases (*"the Fable 5.1 vibe check that we released earlier this week"*) suggest a recurring, recognizable review series. - **Insider speed/access framing:** *"We just got word that it was dropping a few hours ago... Found out at three o'clock in the morning it was dropping."* ## Observable Presentation Patterns (Demonstrated, Not Claimed) - **Comparative, benchmark-style framing:** Models ranked against competitors ("it competes with Anthropic's Fable") and tiered verdicts (*"absolute S-tier daily driver"*). - **Institutional authority through named team members:** e.g., *"our head of video, Randy"* and *"our editor, Jack Cheng"* — demonstrating a real editorial operation. - **Hands-on, artifact-based evidence:** demonstrations include a self-written vibe check, a Waterloo 3D reconstruction, side-by-side app UI comparisons, and a Premiere video edit. - **Balanced review structure:** strengths first (writing, computer use, 3D), then explicit weaknesses (*"now we're getting into some of the things that aren't quite as good"*), ending with a differentiated recommendation by price and task type. - **Casual, personality-forward tone:** *"this vibe check brought to you by Celsius"* — sponsored energy framing within a conversational style. ## Distinguishing Claims vs. Demonstration - The channel's reach and quality (30-person team, 25,000-view prior video) are **creator assertions**, while the comparative testing methodology and verdict structure are **demonstrated** in the video itself. - No channel-wide conclusions (e.g., typical upload cadence, full audience demographics) can be drawn from this single video. # Delivery Observations: "We tested OpenAI's Astra! 5 things to know" ## Overall Structure & Teaching Habits - **Numbered list framing.** The delivery is explicitly organized into countable takeaways: *"Number one, it's a fantastic writer"*, *"All right, number two, it is a beast at computer use"*, through *"Number five"*. Transitions use spoken signposts: *"All right, number three"*, *"Okay, next, number four."* - **Demonstration-first teaching.** Each claim is immediately backed by a concrete artifact the speaker shows or describes — the model's self-written vibe check, the Waterloo 3D reconstruction, the Premiere editing job, the two welcome screens. - **Contrast as an evaluation method.** The criticisms are delivered via head-to-head comparisons: *"Here's the welcome screen from Fable... Here's the welcome screen from Astra"* and *"you can see what I'm saying about the intuitive adherence to the sense of the prompt."* - **Balanced-verdict pattern.** Criticism is consistently paired with a concession: *"but Astra itself has good design taste"*, *"It's got a more well thought out, well designed interface... But it sort of forces you to do one page at a time."* ## Recurring Vocabulary & Phrases - **"Vibe check"** is a signature term, used repeatedly: *"this vibe check brought to you by Celsius"*, *"It one-shotted its own vibe check"*, *"the Fable 5.1 vibe check"*. - **"Bells and whistles"** appears twice, both in criticism contexts: *"It's just extra with all the bells and whistles it puts into interfaces"* and *"It tends to add more bells and whistles."* - **"Extra"** as a standalone judgment: *"Its interfaces are a little bit extra."* - **Day-one urgency framing**: *"a review on day one"*, *"about Astra on day one"*, *"which models to use and for what on day one."* - **Model-jargon shorthand**: *"one-shotted"*, *"S-tier daily driver"*, *"model class"*, *"AI-isms"*, *"slop."* ## Tone - **Casual, conversational register** with interjections and self-talk: *"I just, I gotta show you this"*, *"Like, this is crazy"*, *"it had him going, whoa."* - **Enthusiasm expressed in superlatives**: *"a beast at computer use"*, *"Truly, there's something special happening here"*, *"It is my constant companion."* - **Second-person direct address**: *"Here are five things that you need to know"*, *"you need to check it out"*, *"this thing can do a lot for you."* - **Wonder at the moment**: *"it's just wild that we live in a time where you can just make stuff like this."* ## Promotional Patterns - **Repeated CTA to every.to**, stated at least twice in nearly identical terms: *"you should go to Every, every.to"* and *"Check it out now at every.to"*, followed by another reference: *"live right now on the Every website, every.to."* - **Team-scale credibility claims**: *"a 30-person team taking it through everything from writing, to coding, to design"*, *"extensive testing."* ## Stable vs. Isolated - **Stable-looking patterns**: numbered structure, show-then-judge demos, "vibe check" vocabulary, paired praise/concessions, and the every.to callouts. - **Possibly isolated**: the *"found out at three o'clock in the morning"* anecdote and the *"brought to you by Celsius"* sponsor line appear once each; whether they recur across videos cannot be determined from this single transcript. Similarly, the specific comparison vocabulary ("extra," "bells and whistles") is concentrated in one section, so its status as a general habit is uncertain. # Observed Positions: "We tested OpenAI's Astra! 5 things to know" ## Speaker Attribution The transcript contains a single unnamed host/narrator presenting on behalf of a **30-person testing team at Every**. No individual names are given for the narrator; the only named individuals are colleagues cited secondhand: - **Randy** — head of video, whose computer Astra ran on - **Jack Cheng** — editor, whose prompt comparison is discussed All positions below are attributed to the host unless otherwise noted, and represent *expressed claims from testing*, not independently verified facts. --- ## Claim 1: Astra is a "fantastic writer" - The model "one-shotted its own vibe check" — the host reads the model-generated review of itself, which characterized Astra as "a show horse" with "striking visual taste" whose "ambitions still outrun its judgment." - The host notes the self-review is "a little harsh on itself, but it's, this is good." - Personal assessment: > "It is my constant companion. I feel like its sentences are crisp. They're to the point. There's no AI-isms. There's no slop." ## Claim 2: "A beast at computer use" - Claim: "We've passed this point where it can basically use the computer for you." - Cited evidence: Astra spent ~5 hours in Premiere cutting the *Fable 5.1 vibe check* video (~25,000 views), running on Randy's computer. - Generalization offered: slide decks, video production, spreadsheets — "it multiplies what you're able to do in a day." ## Claim 3: Very good at 3D games and visualizations - Host cites a Battle of Waterloo reconstruction Astra built in "three or four hours," described as historically accurate based on battle accounts and geography. ## Claim 4: Frustrating interface tendencies (qualification begins) - **Positioning:** Astra is a new model class "above Sol" in OpenAI's lineup, competing with Anthropic's **Fable**. - In the team's opinion, Astra has "a few frustrating characteristics" that "keep it from reaching quite Fable's top end." - Specific issue: interfaces are "a little bit extra" — e.g., a login page with unnecessary labels, buttons, and a "no account or sign-in service provided" message. The host notes this happens "a lot, especially at higher effort levels." - Important qualification preserved: **Astra itself has good design taste**; Fable's first take is just "cleaner and more refined." ## Claim 5: Weaker grasp of a prompt's underlying intent - Claim: Astra's adherence to "the underlying sense of your prompt" and extending it in a "delightful, surprising, and still simple" way is "just not as good as Fable." It "tends to add more bells and whistles" and "maybe slightly misunderstand the sense of your prompt." - Evidence (via Jack Cheng's journal-digitization app request to both models): - **Fable's workflow:** simple — show a page, press one button, continuous transcription. - **Astra's workflow:** better-designed interface (warm paper, green highlights) but forces one page at a time with multiple clicks and saves — "not built as intuitively for his use case." --- ## Overall Verdict - **"Absolute S-tier daily driver, if you can afford it."** Priced like Fable, since it's a class above Sol. - Host's personal usage: uses it "every single day for pretty much all my work." - **Explicit reservation preserved:** "for the top end, long running, big delegation tasks, I still use Fable." --- ## Contextual Notes - The review was produced on day one, after learning of the release "at three o'clock in the morning." - The host explicitly promotes Every's subscription, newsletter, courses, and tools (Cora, the Every agent), and points to a fuller "several thousand word" vibe check article — the review is a summarized, promotional framing of the team's testing. - A tension left unresolved in the source: the model is praised as an everyday workhorse while simultaneously judged inferior to Fable for the highest-end tasks; the host holds both positions without reconciling them beyond the pricing/daily-driver framing. # Video Attribution & Provenance Record **Video ID:** 01a0f15a-46a2-7c11-8745-a499c729a305 **Title:** "We tested OpenAI's Astra! 5 things to know" --- ## Speakers & Roles | Speaker | Role | Basis | |---|---|---| | **Main narrator (unnamed)** | Creator / host, presenting first-person testing results | First-person narration throughout ("We've been testing it extensively," "I had it make me this reconstruction") | | **Randy** | Head of video (mentioned third party, not speaking) | Referenced as computer owner for Astra's Premiere editing test | | **Jack Cheng** | Editor (mentioned third party, not speaking) | Referenced as source of the journal-digitization app comparison | | **GPT-5.6/GPT-6 Astra** | AI model output quoted (not a human speaker) | Narrator reads its self-written "vibe check" aloud | > Note: The transcript alternates between "GPT-5.6 Astra" and "GPT-6 Astra" — recorded as-is; discrepancy not resolved by any metadata. No named human narrator identity is provided in the transcript; no curated metadata was supplied to confirm who the host is. --- ## Referenced Sources & Claims - **Internal testing:** A "30-person team" at Every tested the model across writing, coding, and design (creator's own claim). - **Full written review:** A "several thousand word" vibe-check article on every.to (self-referenced source). - **AI model output:** Astra's self-written review excerpt, quoted in transcript. - **Third-party products referenced:** OpenAI (Astra, Sol lineup position), Anthropic (Fable), Adobe Premiere, *Hadestown*, Battle of Waterloo historical accounts, Fable 5.1 vibe-check video (~25,000 views claimed). - **Third-party claims:** Randy and Jack Cheng reactions/results are relayed by the narrator; their views are not directly quoted beyond the narrator's paraphrase. --- ## Promotional Segments — Flagged - **Every subscription promotion** (opening segment): "If you want more of this, you should go to Every, every.to… the only subscription you need to stay at the edge of AI." Promotes newsletter, courses, camps, and internal tools (Cora email agent, Every agent). - **Sponsored mention:** "this vibe check brought to you by Celsius" — explicit sponsorship acknowledgment. - **Every.to repeated call-to-action:** Repeated prompts to visit every.to for the full article and subscription. --- ## Availability Limits - Full testing article available at **every.to** (behind subscription framing: "bundled together under one subscription"). - No mention of comments availability or restrictions. - No transcript-supplied metadata regarding publication date or platform. It's model release day. GPT-5.6 Astra, the long-awaited Astra is dropping today. We just got word that it was dropping a few hours ago. We've been testing it extensively, so we're putting this together for you to have a review on day one. Found out at three o'clock in the morning it was dropping, so this vibe check brought to you by Celsius. Here are five things that you need to know about Astra on day one. From our extensive testing, we've got a 30-person team taking it through everything from writing, to coding, to design, and if you want more of this, you should go to Every, every.to. It's the only subscription you need to stay at the edge of AI. If you wanna be at the edge, you need education and you need equipment, Every provides both. We've got a daily newsletter, we've got courses, we've got camps, we've got everything that you need to know which models to use and for what on day one. And we've got equipment, we build our own tools from an email agent like Cora to the Every agent, an agentic coworker that lives in your Slack. It's all bundled together under one subscription. Check it out now at every.to. And if you really want more, we have the full vibe check. It's a several thousand word article with all of our testing from our whole team live right now on the Every website, every.to. Now, let's get into it. Number one, it's a fantastic writer. It one-shotted its own vibe check. I just, I gotta show you this. Like, this is the vibe check that it wrote for itself based on looking at our Slack and deciding what we thought. Vibe check. GPT-6 Astra is a show horse. OpenAI's new model has striking visual taste and a knack for consulting work. Its ambitions still outrun its judgment. It's a little harsh on itself, but it's, this is good. GPT-6 Astra is very good at getting carried away. OpenAI's new model out today builds beautiful things. I asked her for a 3D rendering of the Hadestown set and got a little theater with weathered green walls, hanging lamps, and a wooden stage ringed with lights. That's, that's very good. That's very hard to do. This model, I've been using it for a while, and I love it for writing. Uh, it is my, it is my constant companion. I feel like its sentences are crisp. They're to the point. There's no AI-isms. There's no slop. It's a very, very impressive writing model, and you need to check it out. All right, number two, it is a beast at computer use. Truly, there's something special happening here with computer use. We've passed this point where it can basically use the computer for you. Uh, a great example, it spent like five hours in Premiere earlier this week, cutting together the Fable 5.1 vibe check that we released earlier this week. That video has like 25,000 views. That, it was cut mostly by Astra, just running on our head of video, Randy's computer, and it had him going, whoa. So, for pretty much anything you're doing on your computer, from slide decks to video production to spreadsheets, this thing can do a lot for you, and it multiplies what you're able to do in a day. All right, number three. It is really good at 3D games and visualizations. I had it make me this reconstruction of the Battle of Waterloo that I can like go around in, and like you can see the British, you can see the French, you can press play. Like, it's all historically accurate from like reading accounts from the battle, and like looking at the geography, and this is, this is crazy. Like, this took, I don't know, three or four hours, and it's just wild that we live in a time where you can just make stuff like this. Incredibly good at 3D visualizations and games. Okay, next, number four. And now we're getting into some of the things that aren't quite as good about this model. Just to take a step back, it's a new model class. It's above Sol in OpenAI's lineup, which means it competes with Anthropic's Fable. And we've found in our testing that it has a few frustrating characteristics that, in our opinion, keep it from reaching quite Fable's top end. Its interfaces are a little bit extra. So like, a very simple example is, I had it put together, a website that had a login page. And you can see this, like, it puts private case study, and then enter shared password to continue, and then password, and then open the case, and then no account or sign-in service provided. Like, that maybe doesn't seem so egregious, but it happens a lot, especially at higher effort levels. And it puts this kind of text and labels and buttons in a lot of different places that don't make as much sense. So what you're gonna find is that Fable's first take on things is just a little bit cleaner and more refined, but Astra itself has good design taste. It's just extra with all the bells and whistles it puts into interfaces. Number five is its understanding and adherence to the underlying sense of your prompt, and then sort of taking it beyond the prompt in a way that is delightful, surprising, and still simple is just not as good as Fable. It tends to add more bells and whistles, and it tends to maybe slightly misunderstand the sense of your prompt. So I'll give you a specific example. Our editor, Jack Cheng, asked both Astra and Fable to create an app for him to help him digitize his handwritten journals. Here's the welcome screen from Fable. You can see it's like pretty simple. Here's the welcome screen from Astra. You can see there's more design taste, but it's got all these buttons. It's got all this stuff. You gotta type all this stuff. It's a little harder to use. But I think the more important thing is, if you look at the process, that Fable designed versus Astra designed, you can see what I'm saying about the intuitive adherence to the sense of the prompt and then extending it. So what Fable designed, pretty simple workflow, just show it a page, press one button, and then go to the next page, and it'll just continuously transcribe. Astra, it has a more well thought out, well designed interface. Like it's got the warm paper, it's got like the green highlights and all that kind of stuff. But it sort of forces you to do one page at a time, click a few buttons, save it, then go to the next one. It's not built as intuitively for his use case. And that's just the thing we find with Astra that keeps it from, in our view, taking the top spot in terms of this big powerhouse that's gonna go for a long time and give you something really usable at the end. So overall verdict, absolute S-tier daily driver, if you can afford it. Remember, this is a model class above Sol, so it's priced like Fable. If you can afford it, I use this model every single day for pretty much all my work. But for the top end, long running, big delegation tasks, I still use Fable.