Today ainotis Join

My notis

ToolingPublished Top storyAll news from that day

Replit chief Masad says he thinks GPT is better now

Replit chief executive Amjad Masad said in a TechCrunch Disrupt 2026 video that cost, partnership and performance made OpenAI a better partner for now.

By ai notis. An AI newsroom with a human editor. Every fact links to its source. How we work

Share

Check our sources · 13 facts from 4 sources
Everyone Can Build Software. What Happens Now? Replit's CEO Amjad Masad Has an Idea | Disrupt 2026: a video from TechCrunch on YouTubeVideo: TechCrunch on YouTube
watch on YouTube · Video still, linking to YouTube

Key points

  1. Asked about Replit's OpenAI partnership and a Free Mode using Luna, Masad said he thinks "GPT is better now" (16:31).
  2. Masad said the decision was based on cost, partnership and performance, and that OpenAI was "a better partner for now" (16:47).
  3. He said enterprise was about 2% of Replit's revenue last year and about 15% this year, and that more than 50% of this year's revenue is B2B (17:25).

What happened

Asked about Replit's OpenAI partnership and its Free Mode, Masad said "GPT is better now" and pointed to Replit's internal benchmark, Vibe Bench, and said Replit publishes a public benchmark as well. He said the decision was based on cost, partnership and performance, and that OpenAI was "a better partner for now". The captions render the OpenAI models he named as "Solan Astra".

What Replit's own pages say is narrower. Its Free Mode post of 18 August 2026 says Free Mode is powered by OpenAI's GPT-5.6 Luna. Its model selector page says that when the product shows Auto, Replit routes each request to a model, and it does not name a default model.

The same page lists OpenAI and Anthropic models side by side in Power Mode and Max Mode, and says each account's selector is the source of truth. The page shows no date.

On the same stage Masad said enterprise contracts are about 15% of Replit's revenue this year, against about 2% last year, and that more than 50% of revenue is business-to-business. These are his figures; Replit says it is not yet sharing ARR.

What it means for you

Our view

Masad's comments are one executive's view, and he points to Replit's own benchmark and says Replit also publishes a public benchmark.

Replit's pages give a narrower picture: its Free Mode post of 18 August 2026 says Free Mode is powered by GPT-5.6 Luna, and its model selector lists both OpenAI and Anthropic models in Power Mode and Max Mode.

If you use Replit, open your own account's model selector and see which models it offers you, since Replit says that list is the source of truth for your account.

It adds no new facts.

Share this story

Your reaction

We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.

Check our sources

Every sentence above is checked against these 4 sources.

1 Everyone Can Build Software. What Happens Now? Replits CEO Amjad Masad Has an Idea | Disrupt 2026TechCrunch (YouTube) · 6 facts Open the source archived copy

  1. The video is titled "Everyone Can Build Software. What Happens Now? Replit's CEO Amjad Masad Has an Idea | Disrupt 2026" and is on TechCrunch's channel.

    Title: Everyone Can Build Software. What Happens Now? Replit's CEO Amjad Masad Has an Idea | Disrupt 2026; channel: TechCrunch
    cite

  2. Asked about Replit's OpenAI partnership announcement of a Free Mode using Luna, Masad said "GPT is better now" (16:31).

    16:10 ... you had a recent um open AI / 16:14 partnership that you announced um free / 16:16 mode using Luna. ... 16:31 >> I think I think GPT is better now.
    cite

  3. Masad said Replit has an internal benchmark called "vibe bench" and publishes a public benchmark as well (16:33-16:38).

    16:31 ... So we / 16:33 actually have a um internal benchmark / 16:36 called vibe bench. We publish a public / 16:38 benchmark as well.
    cite

  4. Masad said that "consistently Solan Astra perform better on building vibe coded applications from scratch", the automatic captions' rendering of the OpenAI models he named (16:39-16:46). This is his account of Replit's benchmarks. the company says

    16:39 And consistently Solan Astra perform / 16:42 better on building vibe coded / 16:44 applications from from scratch.
    cite

  5. Masad said "it was a decision based on both cost, you know, partnership, and performance. And for all these reasons, we just thought like they were a better partner for now" (16:47-16:57).

    16:47 uh it was a decision based on both uh / 16:49 cost, / 16:51 you know, partnership, / 16:52 and and performance. And for all these / 16:55 reasons, we just thought like they were / 16:57 a better partner for now.
    cite

  6. Masad said enterprise was about 2% of Replit's revenue last year and about 15% this year, and that more than 50% of this year's revenue is B2B (17:25-17:41). He said Replit is not sharing ARR numbers yet (17:21). the company says

    17:21 >> We're not sharing ARR ARR numbers just / 17:25 yet, but in terms of growth, like I / 17:26 would say enterprise was like 2% of our / 17:28 revenue last year. ... 17:33 This year, / 17:34 I would say like / 17:36 15% of our revenue is enterprise / 17:38 contracts, and more than 50% of our / 17:41 revenue is B2B
    cite
2 About ReplitReplit · 1 fact Open the source archived copy

  1. Replit's about page lists Amjad Masad as Founder & CEO.

    cite
3 Replit introduces Free ModeReplit · 18 Aug 2026 · 1 fact Open the source archived copy

  1. Replit's blog post of 18 August 2026, updated 19 August, says Free Mode is powered by OpenAI's GPT-5.6 Luna.

    cite
4 Model selectorReplit docs · 5 facts Open the source archived copy

  1. Replit's model selector page says that when Auto shows in the product, Replit performs intelligent model routing, picking a model per task.

    Intelligent model routing chooses an appropriate model for each task, balancing quality and cost. When you see Auto in the product, Replit is performing intelligent model routing.
    cite

  2. Replit's model selector page names no default model and shows no date on the page; its metadata gives 25 September 2026 as the last change.

    Toned down to what the source says
    Free Mode does not have a selectable model list. The list in your Model selector is the source of truth for your account. (page metadata: dateModified 2026-09-25T09:31:53Z)
    cite

  3. Replit's model selector page lists GPT-6 Sol and GPT-6 Luna Fast alongside Claude Sonnet 5 and Claude Sonnet 4.6 under Power Mode.

    Model / Power Mode / Max Mode: GPT-6 Sol (Power), GPT-6 Luna Fast (Power), Claude Sonnet 5 (Power), Claude Sonnet 4.6 (Power)
    cite

  4. Replit's model selector page lists GPT-6 Astra alongside Claude Fable 5.1, Claude Opus 5.5, Claude Opus 4.8, Claude Opus 5, Claude Opus 5 Fast and Kimi K3 under Max Mode.

    GPT-6 Astra (Max), Claude Fable 5.1 (Max), Claude Opus 5.5 (Max), Claude Opus 4.8 (Max), Claude Opus 5 (Max), Claude Opus 5 Fast (Max), Kimi K3 (Max)
    cite

  5. Replit's model selector page says the list shown in your Model selector is the source of truth for your account, and that availability varies by rollout, organization settings and authorization.

    The list in your Model selector is the source of truth for your account. Model selection is rolling out, and the available models can vary by rollout, organization settings, and authorization.
    cite

Topics

The morning email

On the mornings we publish: the three top stories and up to four short ones. Free.

We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep