Experiment: When AI guesses the price of bitcoin

The editorial team at Bitcoin.com News approached eleven leading AI chatbots with the same challenge: set the closing price of bitcoin on 31 December 2026 and justify it. Models from OpenAI, Anthropic, xAI, Google, Microsoft and the Chinese DeepSeek took part in the experiment. Each system received identical context — historical highs and lows, ETF flow developments and current macroeconomic data — and had to justify its prediction in two to three sentences.

Bitcoin is currently trading at around $76,000 to $77,000, which is significantly below the October 2025 high of $126,272. Although the cryptocurrency has partially recovered from the February low of $59,930, caution prevails in the market. It is in this environment that AI models sought an answer to the question of whether a six-figure amount is realistic by the end of the year.

From DeepSeek to Gemini: Who sees what price

Individual estimates span a fairly wide range, but most models agree on mild to moderate growth compared to the current level:

• DeepSeek (Deepthink): $84,500 — the most cautious estimate based on a consolidation phase after the halving.
• Copilot: approximately $92,000 — according to Microsoft's model, the market will not repeat the record levels of 2025.
• Venice AI: $94,500 — a forecast based on typical consolidation 6–9 months after the cycle peak.
• Grok: $108,500 — the xAI model emphasises positive ETF flows and institutional accumulation.
• Gemini 3 Fast: $114,500 — Google's most optimistic scenario relying on monetary easing in the US and global liquidity.

From the other models — Claude Sonnet 4.6, Claude Opus 4.6, ChatGPT 5.3 Instant, Qwen, Pi AI and Le Chat — the editorial team published screenshots whose precise numerical estimates are not quoted in the article. The resulting range of all eleven models moves between $84,500 and $118,400, with most clustering around the $94,000–$110,000 level.

What the models considered: Halving, ETFs and macro liquidity

More interesting than the numbers themselves is the way the models arrived at them. AI systems worked with the same data, but their conclusions differed according to how they weighted individual variables. Among the most frequently mentioned factors were:

The four-year halving cycle

Bitcoin goes through pre-programmed reward halvings for miners, which typically trigger long-term growth and correction cycles. Models such as Grok or DeepSeek emphasised that the current phase corresponds to a typical mid-cycle consolidation, when the market absorbs selling pressure and prepares the ground for later growth.

Institutional ETF flows

One of the key differences compared to previous cycles is the massive presence of spot bitcoin ETFs, which in 2025 and 2026 are influencing demand for the cryptocurrency. The models noted that while ETFs experienced occasional outflows in the first quarter of 2026, flows turned positive again in April. This very factor makes the current market different from previous bearish trends.

Macroeconomic conditions

Gemini and other models also factored monetary easing in the US and global liquidity into their considerations. According to this scenario, bitcoin could benefit from central banks gradually easing their restrictive policies. On the other hand, more cautious models warned that geopolitical tensions and high interest rates could curb growth.

AI predictions vs. reality: Why take the results with a pinch of salt

While the experiment offers a fascinating insight into how different AI systems process market data, it is necessary to remember one fundamental thing: language models are not financial advisors. Their predictions stem from statistical patterns in training data, not from a real ability to predict the future. As shown by the wide range of estimates — almost $34,000 between the lowest and highest result — each model places different weight on different factors.

An interesting comparison is prediction markets. According to data from Polymarket, there is currently an 87% probability that bitcoin will surpass the $80,000 mark by the end of the year, and a 40% chance it will reach $100,000. Although these figures partially overlap with AI estimates, it is still a speculative market driven by emotions and unexpected events that no model can fully capture.

Czech context: Cryptocurrencies and AI tools within reach

For Czech readers, the experiment offers at least two practical lessons. First, all the mentioned AI tools — ChatGPT, Claude, Gemini and Copilot — are commonly available in the Czech Republic, although most of them primarily communicate in English. Czech is best supported by ChatGPT and Google Gemini; for specialised financial queries, however, English is still recommended for more accurate results.

Secondly, trading in cryptocurrencies is legal in the Czech Republic and subject to European Union regulations, particularly the MiCA (Markets in Crypto-Assets) regulation, which entered into force in 2024. The Czech National Bank has repeatedly warned that investments in cryptocurrencies carry a high risk of loss, and its predictions — whether from AI or analysts — should not serve as a basis for investment decisions without one's own research.

Conclusion: AI opens the discussion, it does not close the forecasts

The experiment with eleven AI models did not provide a clear answer, but something perhaps more important: it showed how different artificial intelligence systems interpret the same data through different "lenses". While some models see cautious consolidation, others see a path to six-figure numbers. The common denominator is the belief that bitcoin will likely exceed its current level by the end of 2026, without however repeating the records from autumn 2025.

For technology enthusiasts and investors alike, it remains key to realise that AI is a tool for structuring information, not a crystal ball. Whether bitcoin stands at $85,000 or $115,000 on New Year's Eve, the investment decision should always rest on a solid own risk assessment — not on a tip from a chatbot.