# Comparison Articoli collegati

Il Centro Notizie HTX fornisce gli articoli più recenti e le analisi più approfondite su "Comparison", coprendo tendenze di mercato, aggiornamenti sui progetti, sviluppi tecnologici e politiche normative nel settore crypto.

Same $5 Rate, Bill Differs by 30%, OpenAI Exec: Token Pricing Is Never Directly Comparable

Here is a summary of the article in English: **Title: Priced at $5, Bills Vary by 30%. OpenAI Executive: Token Prices Are Not Directly Comparable** A key takeaway from OpenAI's Codex lead, Tibo, is that a "token" is not a standardized unit for comparing AI model costs, akin to grams or kilowatt-hours. He uses an analogy: two identical pizzas priced per slice can yield different total costs depending on how they're cut. Similarly, different models use different "tokenizers" to segment text, meaning the same input text can produce vastly different token counts. For instance, the same text was tokenized as 766 tokens by GPT-5.6 Sol and 1170 tokens by Claude Opus 5—a 34.5% difference—despite both models advertising the same input price of $5 per million tokens. This discrepancy arises because each company trains its own tokenizer based on its training data, affecting how common or rare word combinations are split. The problem isn't cross-vendor only. Even Anthropic warns that its newer models (Claude 4.7+) use a different tokenizer, producing roughly 30% more tokens for the same text than earlier versions, so cost estimates shouldn't be reused across model generations. Bill differences stem from four main factors: 1) Tokenizer efficiency (input token count), 2) Caching (e.g., GPT-5.6 Sol offers a much lower cache input rate), 3) Output pricing (which can outweigh input savings in agent workflows), and 4) Context length pricing tiers (e.g., GPT-5.6 Sol charges double the input rate for entire requests exceeding 272K tokens). Tibo also addressed user reports of GPT-5.6 Sol's context window being limited in practice, sharing configuration code to manually expand it to 1 million tokens. However, he cautioned that larger windows increase the risk of hitting higher pricing tiers as longer session histories are processed repeatedly. The article concludes that the true metric should shift from "price per million tokens" to "price per successful outcome." The most accurate way to compare costs is to run identical real-world tasks (with the same prompts, tools, and data) through different models, accounting for all variables like tokenization, caching, output length, and context pricing. Ultimately, what matters is the total cost to complete a specific job, not the nominal token price.

marsbit08/19 08:21

Same $5 Rate, Bill Differs by 30%, OpenAI Exec: Token Pricing Is Never Directly Comparable

marsbit08/19 08:21

Bitcoin Hits Six-Year Low Against Stocks. What This Means

Bitcoin has updated a six-year negative record relative to stocks, according to analytics platform Glassnode. Over the last three months, Bitcoin ($BTC) has outperformed the S&P 500 index on only about one-third of trading days, marking its lowest success rate (37.8%) in six years. This contrasts with 2023-early 2024 when it outperformed 50-60% of the time. For the year through August 18, Bitcoin has fallen nearly 27%, while the S&P 500 and NASDAQ have risen 14% and 19%, respectively, and gold gained almost 2%. This underperformance level is comparable to early 2022, just before the start of a prolonged crypto bear market. While the 2026 downturn hasn't seen an event like the FTX collapse, experts note it is one of the toughest periods, with multiple exchanges closing and miners shifting to AI services. Longer-term, from 2021 onwards, Bitcoin's gain of about 120% is closer to traditional assets like gold and the NASDAQ (both up over 130%) and the S&P 500 (up ~105%). Analysts note Bitcoin's historical excess returns are diminishing, though it still shows strong growth rates overall. The current cycle since 2024 is seen as positive for institutional adoption due to Bitcoin ETF approvals and new US regulatory approaches.

cryptonews.ru08/18 14:09

Bitcoin Hits Six-Year Low Against Stocks. What This Means

cryptonews.ru08/18 14:09

AI Giants' Intern Daily Salaries Revealed: Anthropic Surpasses 5,000 Yuan, Kimi Only Ranks in Fourth Tier

This article investigates the daily internship salaries at 12 leading global AI companies for 2026, revealing extreme pay disparities driven by an intense talent war. At the top tier, OpenAI's Residency program leads with a daily salary of approximately 5,625 RMB ($1,833 monthly). Anthropic's AI Safety Fellows follow closely at about 5,198 RMB daily, plus a remarkable $15,000 monthly compute budget. Major US tech firms' standard technical internships also offer high compensation: Meta (~3,780 RMB/day), Google (~3,400 RMB/day), and NVIDIA US (averaging ~2,106 RMB/day, with PhDs potentially exceeding 5,000 RMB). Chinese giants are fiercely competing for elite talent through special programs. ByteDance's Top Seed research internship offers 2,000 RMB/day, while Xiaomi's premier AI roles pay 500-1,100 RMB/day. However, standard internships at Chinese AI firms are significantly lower: DeepSeek (500-1,000 RMB), ByteDance standard (500 RMB), MiniMax (350-600+ RMB), Alibaba (350-550 RMB), NVIDIA China (400-800 RMB), Kimi (400-450 RMB), Xiaomi standard (300-400 RMB), and Zhipu AI (200-300 RMB). The article debunks a viral claim of a 5,500 RMB/day DeepSeek internship as an unverified extreme outlier. Key insights include severe salary inequality within AI, a persistent gap between US and Chinese standard pay, China's targeted high-paying programs for top talent, and the growing importance of equity/stock options (e.g., at Zhipu, MiniMax, Kimi) alongside cash compensation. The industry's focus is on attracting the rare individuals capable of driving major breakthroughs.

Odaily星球日报08/15 04:31

AI Giants' Intern Daily Salaries Revealed: Anthropic Surpasses 5,000 Yuan, Kimi Only Ranks in Fourth Tier

Odaily星球日报08/15 04:31

Opus 5 Burns 690M Tokens Making a Game; GPT-5.6 Replicates It for $5

A user used a single, highly detailed 2000-word prompt with the Opus 5 model to generate a complete browser-based 3D jet ski racing game called "INK TIDE," spending $423 and consuming 6.9 billion tokens. The prompt acted as a comprehensive project brief, specifying technical constraints, visual style, and a multi-agent workflow to prevent "attention dilution" by dividing tasks like water simulation, physics, and AI among specialized sub-agents. The resulting game features cartoon-style graphics, a functional UI, and basic AI opponents. Subsequently, another user replicated a similar game using the GPT-5.6 Sol and GPT-5.6 Luna models within Codex, costing only approximately $5. While the $5 version captures the core racing concept and aesthetic, its visual fidelity and polish are notably less refined than the original Opus 5-generated demo. The comparison highlights that while AI can now produce technically functional games, the final quality and cost are heavily influenced by the model used, the detail of the human instruction, and the allocated budget.

marsbit08/10 00:31

Opus 5 Burns 690M Tokens Making a Game; GPT-5.6 Replicates It for $5

marsbit08/10 00:31

Is Using an iPhone as a Cryptocurrency Wallet Effective?

The article discusses the suitability of using an iPhone as a cryptocurrency wallet amid heightened security concerns, including fears of quantum computing decryption and recent exploits like the Coldcard hardware wallet breach. Security investigator ZachXBT suggested in Telegram that a dedicated, older iPhone could be a viable alternative to specialized hardware wallets, though he noted this approach isn't for everyone. Roman Storm, a key developer of Tornado Cash, acknowledged the idea but highlighted a critical limitation: the lack of a unified mobile wallet application supporting the BIP39 passphrase feature. This optional security layer adds a word or string to a standard seed phrase to create a hidden wallet. Storm argued that a core security threat—someone physically accessing your device at home—remains unaddressed by simply switching to an iPhone. He urged mobile wallet providers to implement BIP39 support. The article notes that dedicated hardware wallets like Trezor, Keystone, BitBox, BitBox, and Coldcard already support BIP39, giving them a current advantage. Representatives from Keystone and Trezor's Danny Sanders added that iPhones, as general-purpose devices, come with more features and potential attack surfaces, requiring significant technical knowledge and strict security discipline to use safely as a wallet.

cryptonews.ru08/03 23:09

Is Using an iPhone as a Cryptocurrency Wallet Effective?

cryptonews.ru08/03 23:09

How Token-Hungry is Claude Code? A Comparative Experiment Shows Up to 30x Difference Across Three Frameworks

Claude Code's Token Consumption Exposed: Comparison Experiment Shows Up to 30x Difference Between Frameworks A recent experiment by the Composio team tested the same model (Kimi K3) across three different agent frameworks (Claude Code, Hermes, and Kimi Code) on 28 identical tasks. While task completion rates were similar, token consumption varied dramatically. The median token usage was approximately 61k for Kimi Code, 67k for Hermes, and a staggering 340k for Claude Code – about 6 times more than Kimi Code. For individual tasks, the maximum difference reached 30x. In terms of cost, using Claude Code averaged $2 per task compared to $0.22 for Kimi Code and $0.28 for Hermes (based on Kimi K3 pricing). Speed also differed, with Hermes being the fastest. Analysis suggests Claude Code's high token usage stems from its harness repeatedly feeding extensive context (previous messages, tool calls, command outputs, file contents) back into the model across multiple interaction rounds, significantly inflating input tokens rather than generating longer outputs. This highlights a crucial trend: the agent framework (harness) is becoming as important as the model itself for cost and efficiency. A separate study from Writer showed that simply switching the orchestration layer to their optimized harness reduced average task cost by 41% and latency by 44% across various models without sacrificing quality. The conclusion is clear: for cost-effective AI agents, optimizing the harness may yield greater savings than changing the model. The future of agent competition may hinge not just on capability ("can it do it?") but on efficiency ("who does it for less?").

marsbit07/31 12:26

How Token-Hungry is Claude Code? A Comparative Experiment Shows Up to 30x Difference Across Three Frameworks

marsbit07/31 12:26

From Gold to Bitcoin: Fixed Supply + Institutional Frenzy, Might It Repeat the 'Explosive' Price Trend?

"From Gold to Bitcoin: Fixed Supply and Institutional Frenzy May Lead to 'Explosive' Price Rally Analysts suggest Bitcoin's price action could mirror gold's over the past two decades, following the launch of spot Bitcoin ETFs. Gold ETFs, introduced in 2004, drove gold's price surge to a current market cap near $28 trillion. Both gold and Bitcoin are non-yielding stores of value, with prices driven purely by investor sentiment rather than cash flows or credit. Gold ETFs experienced dramatic cycles: explosive growth, painful drawdowns, and slow recoveries, with each cycle reaching higher peaks. Bitcoin ETFs, approved in early 2024, saw rapid institutional adoption but are now facing similar volatility. Recent warnings highlight the risk of significant ETF outflows disrupting the current rebound. BlackRock's IBIT, a leading Bitcoin ETF, has sold nearly 100,000 BTC to meet redemptions while still holding over 733,000. The core parallel is fixed supply: when demand surges, prices explode, but demand is often volatile and wave-like, not steady. Institutional interest, through ETFs and corporate adoption, remains a key support pillar, helping to cushion sell-offs. If Bitcoin captures even a fraction of gold's role as a store of value, its upside potential is immense, though the path will be marked by high volatility. For investors, focusing on long-term trends and managing risk is crucial as this 'price explosion' narrative unfolds."

Foresight News07/20 07:03

From Gold to Bitcoin: Fixed Supply + Institutional Frenzy, Might It Repeat the 'Explosive' Price Trend?

Foresight News07/20 07:03

Top 7 VPNs of July 2026

With over 6 billion internet users globally, only a small percentage utilize a VPN to encrypt their connection, mask their IP address, and protect data, especially on public Wi-Fi. Choosing a VPN requires considering speed, privacy record, and server network. Here are the top 7 VPNs for July 2026: 1. **ExpressVPN:** Offers high speed, reliability, and usability with over 3,000 servers in 106 countries, featuring a proprietary Lightway protocol. 2. **Windscribe:** A privacy-focused, freemium VPN committed to an open internet, offering server-side ad blocking and browser extensions. 3. **Norton VPN:** A user-friendly option with a strict no-logs policy, kill switch, and split tunneling, integrated into the Norton 360 security ecosystem. 4. **CyberGhost:** An affordable VPN with a 45-day money-back guarantee, over 12,000 servers, and independently audited no-logs policy. 5. **Mullvad VPN:** Prioritizes privacy by not requiring an email or account, offering flat pricing, strong speeds, and optional content blocking. 6. **NordVPN:** A popular provider with servers in 118 countries, standard protections, and higher-tier plans including Threat Protection Pro. 7. **Proton VPN:** Offers a generous free tier and paid plans with features like Secure Core and NetShield, operating under strict Swiss privacy laws. VPNs are crucial for protecting online activity. Users should research thoroughly before purchasing any service.

ambcrypto07/07 16:53

Top 7 VPNs of July 2026

ambcrypto07/07 16:53

First Batch of GPT-5.6 Sol Internal Test Results Are In, Same-Task Cost is Only Half That of Fable 5

"GPT-5.6 Sol", the preview version of OpenAI's latest model, has released its first batch of internal testing results. According to an Nvidia engineer, Sol achieved CUDA acceleration results in 30 hours that took Opus 64 hours. The model is positioned for high-difficulty reasoning and complex, long-chain tasks like coding. Initial user feedback highlights that Sol produces more concise code—approximately 1/5 the lines of Opus—with a style closer to human-written code, favoring deep optimization over brute-force trial and error. In tests, Sol demonstrated superior performance in instruction following, spatial reasoning, and front-end design consistency compared to GPT-5.5 Pro. When compared to the recently relaunched competitor Fable 5, Sol shows slightly lower overall performance and code quality in some areas. For instance, Fable 5 can generate a playable game from a single prompt faster. However, Sol's key advantage is cost: at $5 per million input tokens and $30 per million output tokens, it is roughly half the price of Fable 5 ($10/$50). Additionally, users report Fable 5's safety restrictions are overly stringent, frequently blocking tasks, whereas Sol's safeguards are reportedly more tailored and less limiting. The model is expected to roll out fully to all users in the coming days.

marsbit07/06 07:29

First Batch of GPT-5.6 Sol Internal Test Results Are In, Same-Task Cost is Only Half That of Fable 5

marsbit07/06 07:29

To Buy Discounted ETH, How to Choose Between Bitmine and SharpLink?

"Buying Discounted ETH: SharpLink vs. Bitmine" analyzes two major treasury companies during a period of low ETH prices. Both hold significant ETH positions at a loss: SharpLink (with ~886.7K ETH at an average cost of ~$3,609, unrealized loss >$1.7B) and Bitmine (with 5.7M ETH, ~4.7% of supply, cost ~$3,400, loss >$11B). Despite similar ETH price declines and inclusion in Russell indices, their market valuations differ markedly. SharpLink trades at a ~21% discount to its ETH net asset value (NAV), while Bitmine's discount is only ~6%. The article compares them across key dimensions: financing ability (Bitmine raised $19.2B aggressively vs. SharpLink's smaller, steady raises), stock liquidity (Bitmine is significantly more liquid), and narrative execution (both support Ethlabs; SharpLink plans RWA tokenization, yet to materialize). For short-term trading on an ETH rebound, Bitmine offers better liquidity and a smaller NAV discount. For long-term holding, SharpLink's deeper discount and simpler capital structure (vs. Bitmine's perpetual preferred shares) may offer more upside if ETH recovers, though it carries higher liquidity risk. The choice depends on the investor's risk preference between SharpLink's visible price/ liquidity weakness and Bitmine's hidden risks in its capital structure and past valuation premiums.

marsbit07/03 06:48

To Buy Discounted ETH, How to Choose Between Bitmine and SharpLink?

marsbit07/03 06:48

活动图片

Othersright-arrow

Trading Strategiesright-arrow