The short answer
Pick Claude for coding, for turning spreadsheets and PDFs into dashboards, and for work where quality matters more than cost. Pick Gemini when you need fast responses at a low price, when your team lives in Google Workspace, or when you want to build an app with AI inside it in Google AI Studio.
We use both, but not equally. Through most of 2025 and 2026, Anthropic's models were our default for building, and Gemini was the one we kept testing. Jordan's January 2026 line sums it up: if Gemini's tools "become top notch, you know, we probably move over to there."
What we built with each
Claude
- CFO dashboard from Google's 10-K. Claude turned a 57-page PDF into a full dashboard in one go (Episode 11).
- Cholesterol medication 3D visualizer: built only by chatting, because Claude has "a very good web rendering engine" (Episode 13).
- ScreenEval: "Opus 4.5 for everything" (Episode 29).
- Kalshi prediction market trading bot and the Claude-generated n8n outreach workflow.
Gemini
- Collective finance dashboard: Jordan moved to Gemini because the P&L and balance sheet data was too big for the smaller model he started with (Episode 14, at 06:59).
- Mirror, Mirror on the Wall, Guess My Costume and the Trick-or-Treat route planner: Gemini built into AI Studio did all the AI work (Episode 19).
- AI start-of-care packet generator: Gemini Pro in AI Studio, about thirty minutes (Episode 24).
Both
The LLM Math Roaster ran Gemini 2.5 Pro, GPT-5, Claude Sonnet 4.5 and Grok 4 Fast on the same proof. Gemini got the best score, 98 (Episode 22, at 08:15). It was one problem, so we do not read much into it.
Head-to-head
The sharpest cost-versus-quality take came from David Petrou in September 2026, right after Fable 5.1 shipped:
Still, though, when you look at cost and trying to be, you know, Pareto optimal on on cost versus quality, things like Gemini 3.8 Flash, which just came out today, it looks really good.
— David Petrou, Episode 51
He credited Anthropic with fixing Fable 5's "loquacious" and "oblique" language in 5.1, but argued that more tokens per dollar may matter more than the absolute frontier.
Jordan's reason for leaning on Anthropic in early 2026 was coding. In Episode 29 he said "as long as Anthropic coding models continue to be the best, it seems like coding is everything," because the new coworker-style tools are built on top of the coding models. He named Gemini 3.5 as one of the releases he was watching.
Where each one fell short
Claude. Price and surprise bills. Arun Kalaiselvan said Arya Health's bill once came in at six times what they expected, after a change in how long their Lambda held the connection to Anthropic open (Episode 24, at 14:33). In Episode 46 Jordan called Fable "very slow" and said it "seems expensive." David also noted Anthropic makes subscription users go through Claude Code rather than letting them plug the subscription into other tools.
Gemini. Inconsistent quality. Jordan said in Episode 27 that Gemini "hasn't been a top tier model yet compared to some of the other models on the market" in some places. On Gemini 3's launch day Jordan got throttled in Cursor and Antigravity, and Sam hit "some hiccups" building a meditation app.
What our guests use
- August Kiles, Emblem: "I think we probably use Claude and Gemini the most in our actual products," plus Grok and OpenAI (Episode 31, at 12:35). He called Opus 4.6 "truly an inflection point."
- Anand Chandrasekaran, Arya Health: Claude via Bedrock, Gemini via Vertex AI for HIPAA workflows; Gemini for speed, and newer Claude versions turning multi-shot prompts into single-shot.
- Sagi Waitzman: Claude for the research team, Gemini for marketing posts because it comes with Google Workspace.
- Eldar Sadikov: Anthropic's enterprise plugins are "a logical step," while Gemini already had an advantage plugging into Gmail and Calendar (Episode 33).
- Carina, Axiom: "We like Gemini three," mostly for its price.
- Ben Lerner, Espresso AI, who once worked on language processing at Google: "I think all of these models are keeping pace with each other," and in any given month one might be a little ahead (Episode 27, at 18:22). He was excited to see Gemini power Siri.
Which one we reach for now
Claude for anything we will ship or analyze, Gemini inside AI Studio for fast prototypes. In Episode 43 (May 2026) Jordan said Google's TPUs and low token costs "could provide for a pretty dangerous recipe" if it gets a really good model on top, and he planned to test Gemini 3.5 Flash. By October 2026 he was testing Anthropic's new Opus, which he said was "performing quite well and significantly cheaper than using Fable" (Episode 53). For the OpenAI side, see Claude vs ChatGPT and Gemini vs ChatGPT.
FAQ
Is Claude better than Gemini?
For coding and turning documents into dashboards, it has been our pick for most of the show. Gemini scored highest in our LLM Math Roaster test, though, and Jordan was impressed with Gemini 3's coding and design taste at launch.
Is Gemini cheaper than Claude?
We did not run a price test. Guests say yes: Carina of Axiom called Gemini 3 much cheaper to run, Anand Chandrasekaran called Claude Code already very expensive, and David Petrou said Gemini 3.8 Flash looks really good on cost versus quality.
Is Gemini or Claude better for coding?
Claude, in our experience and most guests'. Anand said in December 2025 that Gemini 3 can do some coding work but is still not there yet. Jordan said in January 2026 that Anthropic models were the best way to build, no doubt.
Do companies use Claude and Gemini together?
Yes. Emblem uses Claude and Gemini the most in its product, and Arya Health runs Claude through Bedrock and Gemini through Vertex AI.