Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
81 76 87 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
Insufficient record: 2 predictions resolved, 5 needed for a score.
-
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
65 45 83
All types combined: 80 (likely 75 to 85, high confidence), from 84 claims: 78 factual claims, 6 promises. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Demis Hassabis's claims
1 claim
| Verdict | Kind | Said | Claim | Topic |
|---|---|---|---|---|
| Mostly false | Factual | 20 May 2026 | Gemini 3.5 Flash is often at less than half the cost of comparable models. Factual · 20 May 2026 | Economics and markets |
“Often at less than half the cost”
Demis Hassabis, Google DeepMind CEO, on X 20 May 2026 Post by @demishassabis on X, 2026-05-20 (X)
Gemini 3.5 Flash is often at less than half the cost of comparable models.
Mostly false False in substance, right in a detail.
The second judge would not resolve this claim; the verdict shown is the first judge's.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Judged as of 2026-05-20. Check published per-token API prices for Gemini 3.5 Flash against comparable models from other providers at launch (models of a similar tier or similar benchmark performance, which is the comparison the claim implies). True: in many of the relevant comparisons (for example several comparable models, or on input and output pricing), Gemini 3.5 Flash's price is less than 50% of the comparable model's price. Mostly true: it is clearly cheaper and below half in some comparisons, but 'often' is a slight stretch, for example only a few comparisons fall just under half. Mixed: it is cheaper than comparable models, but under half in only a minority of comparisons, or the result depends heavily on which models are picked. Mostly false: it is cheaper in some comparisons but almost never below half. False: it is usually priced the same as or higher than comparable models, or the model did not exist as described.
Evidence
- Gemini 3.5 Flash (Minimal) vs GPT-5 (High): Model Comparison | Artificial Analysis Artificial Analysis
“Gemini 3.5 Flash (Minimal) is cheaper. At a 7:2:1 cache hit/input/output ratio, Gemini 3.5 Flash (Minimal) costs $1.31 per 1M tokens (based on Google's API), compared with GPT-5 (High) at $1.34 per 1M tokens (based on OpenAI's API).”
- Gemini 3.5 Flash (Minimal) vs GPT-5 (High): Model Comparison | Artificial Analysis Artificial Analysis
“| Input Price per 1M Tokens | $1.50 | $1.25 | |”
- Gemini 3.5 Flash (Medium) vs Gemini 3.1 Pro Preview: Model Comparison | Artificial Analysis Artificial Analysis
“Gemini 3.5 Flash (Medium) is cheaper. At a 7:2:1 cache hit/input/output ratio, Gemini 3.5 Flash (Medium) costs $1.31 per 1M tokens (based on Google's API), compared with Gemini 3.1 Pro Preview at $1.74 per 1M tokens (based on Google's API).”
- Gemini 3.5 Flash (High) vs Gemini 3.1 Flash-Lite: Model Comparison | Artificial Analysis Artificial Analysis
“Gemini 3.1 Flash-Lite is cheaper. At a 7:2:1 cache hit/input/output ratio, Gemini 3.1 Flash-Lite costs $0.22 per 1M tokens (based on Google's API), compared with Gemini 3.5 Flash (High) at $1.31 per 1M tokens (based on Google's API).”
- Gemini 3.5 Flash: The new leader in intelligence versus speed Artificial Analysis · 19 May 2026
“Gemini 3.5 Flash costs $1,552 to run the Artificial Analysis Intelligence Index, 5.5x more than Gemini 3 Flash and 75% more than Gemini 3.1 Pro.”
- Gemini 3.5 Flash (High) vs GPT-5 mini (High): Model Comparison | Artificial Analysis Artificial Analysis · 1 January 2026
“GPT-5 mini (High) is cheaper. At a 7:2:1 cache hit/input/output ratio, GPT-5 mini (High) costs $0.27 per 1M tokens (based on OpenAI's API), compared with Gemini 3.5 Flash (High) at $1.31 per 1M tokens (based on Google's API).”
- Claude Haiku 5.5 vs Gemini 3.5 Flash — Benchmarks Compared AI Release Tracker · 7 October 2026
“Only Claude Haiku 5.5 has a verified first-party price for an available API: $0.10 per million input tokens and $0.50 per million output tokens.”
Reasoning
Artificial Analysis gives Gemini 3.5 Flash a blended price of $1.31 per 1M tokens. That is only slightly below GPT-5 (High) at $1.34 (E1), and its input price of $1.50 per 1M tokens is above GPT-5's $1.25 (E2). Gemini 3.5 Flash is cheaper than Gemini 3.1 Pro Preview by only about a quarter on blended price ($1.31 vs $1.74) (E3), and it cost 75% more than 3.1 Pro to run the Intelligence Index (E5). Artificial Analysis finds GPT-5 mini (High) about five times cheaper on blended price ($0.27 vs $1.31 per 1M tokens) (E6). In none of these comparisons is Gemini 3.5 Flash below half the price, though it is cheaper in some, which matches the "mostly false" criterion rather than "true" or "mixed".
Cite this claim
Demis Hassabis, “Gemini 3.5 Flash is often at less than half the cost of comparable models.” (said 20 May 2026). Verdict: mostly false. Reliability API, claim c_832fa71cfbe5, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_832fa71cfbe5
Claims without a verdict
Is a verdict here wrong, or do you speak for Demis Hassabis? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.