Replit is no longer working with Anthropic, TechCrunch reported on 16 October 2026, citing CEO and co-founder Amjad Masad onstage at TechCrunch Disrupt. Masad gave three reasons: “It was it was a decision based on both cost, you know, partnership, and performance.”
The account was disputed the same day. One person familiar with the partnership, who asked not to be named, told TechCrunch it was still in place, and a Replit blog post from 29 September lists Anthropic’s Fable 5.1 running in Replit Agent production that month.
Business Pill · SWITCHING COSTS
A one-minute explainer of switching costs: why it can be hard to leave an AI supplier. It teaches the general idea only and says nothing about any company in this story.
The key insight: As we read it, the dispute is about where a coding platform’s loyalty sits: Masad says Replit is no longer working with Anthropic and cites cost, partnership and performance, while a person familiar told TechCrunch the partnership is still in place and Replit’s own September post lists Fable 5.1 in production.
What Masad Said
“I think ChatGPT is better now,” Masad said, according to TechCrunch, adding that Replit uses an internal benchmark to assess metrics like performance.
He pointed to Replit’s published results: “We publish a public benchmark as well, and consistently, Sol and Astra perform better on building vibe-coded applications from scratch.” He added: “And for all these reasons, we just felt like they were a better partner for now.”
TechCrunch notes that Replit announced a partnership with OpenAI in August, “but there was never any indication that its was no longer with Anthropic.” Under the Anthropic partnership, it says, Replit used Claude models to run its platform, a partnership that “was said to have helped the company grow its revenue significantly.”

The Pushback
TechCrunch reports that Masad’s comments “were met with some pushback from insiders familiar with the partnership between the Anthropic and Replit.” One person familiar, who asked not to be named, “said the partnership was still in place.”
TechCrunch also points to a September Replit blog post that “cites Anthropic’s Claude Fable 5 and Claude Fable 5.1 as models running in Replit Agent’s production environment.” It says it reached back out to Replit for clarification.
What Replit’s Own Posts Show
Replit’s 18 August post launched Free Mode, which the company says is powered by OpenAI’s GPT-5.6 Luna and lets Core subscribers create 30x more than before for their monthly subscription; it lists the price as $20 per month.
Replit’s 29 September engineering post, “Free the models”, describes a table drawn from “Replit Agent production, one week per model: Fable 5 in August 2026, Fable 5.1 and Astra in September.” In that table, Fable 5, Fable 5.1 and GPT-6 Astra dispatched a subagent on 32%, 21% and 36% of turns.
The same post says “All three models delegate, but each in its own way”, and that “GPT-6 Astra hands routine implementation to less costly subagents and decides for itself where its tokens are worth spending.” Astra handed work to a general worker on 20% of turns, against 0.9% and 2.3% for the two Fable models, per the table.
What Replit Says About Its Business
Masad also gave a growth figure onstage, TechCrunch reports: “We’re not sharing ARR numbers just yet, but in terms of growth, like, I would say, enterprise was like 2% of revenue last year.” He continued: “This year, I would say like 15% of our revenue is enterprise contracts, and more than 50% of our revenue is B2B.”
Model choice is visible elsewhere in the market too. Earlier today we reported that TypeSafe’s Jev reached 35.3% of paid teams on Vercel’s AI Gateway in its first 14 days, and that Claude Opus 5.5 fast mode left preview at $8/$40 per million tokens. Earlier this week we wrote that Claude Code reached 87% of coding-agent buyers on Ramp.
The Structural Read
As we read it, Replit’s own posts show it running several labs’ models side by side: the 29 September post compares Fable 5, Fable 5.1 and GPT-6 Astra in production, one week each, and the 18 August post put Free Mode on OpenAI’s GPT-5.6 Luna.
As we read it, the reasons Masad gave are commercial as well as technical: he names cost and partnership next to performance, and says Replit’s benchmark shows Sol and Astra performing better on building vibe-coded applications from scratch.
As we read it, the two accounts have not been reconciled: TechCrunch says there was never any indication Replit had stopped working with Anthropic, a person familiar says the partnership is still in place, and TechCrunch says it reached back out to Replit for clarification.
Amjad Masad, Replit CEO, at TechCrunch Disrupt (16 October 2026, per TechCrunch)
Three Implications
AI CODING PLATFORMS Replit’s 29 September post describes running Fable 5, Fable 5.1 and GPT-6 Astra in production one week each, and its CEO now says cost, partnership and performance decided its partner.
MODEL LABS SELLING TO PLATFORMS TechCrunch says the Anthropic partnership was said to have helped Replit grow its revenue significantly; Masad says Replit no longer works with Anthropic, which a person familiar disputes.
ENTERPRISE BUYERS Masad says about 15% of Replit’s revenue is now enterprise contracts, up from about 2% last year, and more than 50% is B2B, according to TechCrunch.
The Business Engineer Lens
This story maps onto the Business Engineer framework The Agentic Harness War.
The framework argues: “Foundation models become relatively commoditized as orchestration layers capture differentiation”.
As we read it, Replit’s posts fit that description: its harness, Replit Agent, ran models from Anthropic and OpenAI in turn, and its CEO describes the model supplier as a choice made on cost, partnership and performance. The framework predates the announcement and does not discuss Replit.

What Is Not Established
TechCrunch says it reached back out to Replit for clarification; at the time we read the article it had not added a reply. We found no statement from Anthropic on Masad’s comments.
Neither TechCrunch nor Replit’s posts give contract terms, the date any Anthropic arrangement changed, or whether Anthropic models still run anywhere in Replit’s products.
The Bottom Line
Replit CEO Amjad Masad said onstage at TechCrunch Disrupt on 16 October 2026 that Replit is no longer working with Anthropic, citing cost, partnership and performance, and said Sol and Astra perform better on Replit’s benchmark.
A person familiar with the partnership told TechCrunch it is still in place, and Replit’s own 29 September post lists Fable 5.1 in Replit Agent production in September; TechCrunch has asked Replit to clarify.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. We read in full TechCrunch’s report of Amjad Masad’s onstage remarks at TechCrunch Disrupt (16 October 2026) and Replit’s blog posts of 29 September and 18 August 2026 on 16 October 2026; every quotation and figure comes from those texts. We checked Masad’s X posts of 13 to 16 October and found none on Anthropic. We did not contact Replit or Anthropic and found no Anthropic statement. The references to our earlier reporting link to our own pieces.
Sources: TechCrunch: Replit is no longer working with Anthropic, CEO says (16 Oct 2026) · Replit: Free the models, Harness design at the frontier (29 Sep 2026) · Replit: Replit Introduces Free Mode (18 Aug 2026)








