One upper limit for how many transactions can go into a block is 4,000,000WU (weight units, where 1 byte of witness data accounts for 1WU, while for non-witness transaction data 1 Byte accounts for 4WU. There's no relation between number of transactions in a candidate block and how long it takes to mine a valid blockheader hash. Depending on how many unconfirmed transactions are available in the mempool of a mining pool, a block candidate is commonly (quite simplified) assembled with transactions (or transaction packages in the case of child-pays-for-parents chains, parents must occur before child in block) from highest fee rate to lowest until the block is full (not exceeding 4M WU). This usually gives the mining pool maximum profit from available transaction fees in mempool. There are more limits to pay attention to, like number of Sigops can't exceed 80,000 (there has been at least one broadcasted invalid block exceeding this limit by a small amount, I think last year, big oopsie by certain mining pool, lol). In times when mempool isn't full enough to fill a block with transactions equal and above fee rate of 1sat/vB it makes economical sense to include transactions that pay less than 1sat/vB when those are available in mempool. Not all mining pools do that, so far. Some do...
One upper limit for how many transactions can go into a block is 4,000,000WU (weight units, where 1 byte of witness data accounts for 1WU, while for non-witness transaction data 1 Byte accounts for 4WU. There's no relation between number of transactions in a candidate block and how long it takes to mine a valid blockheader hash. Depending on how many unconfirmed transactions are available in the mempool of a mining pool, a block candidate is commonly (quite simplified) assembled with transactions (or transaction packages in the case of child-pays-for-parents chains, parents must occur before child in block) from highest fee rate to lowest until the block is full (not exceeding 4M WU). This usually gives the mining pool maximum profit from available transaction fees in mempool. There are more limits to pay attention to, like number of Sigops can't exceed 80,000 (there has been at least one broadcasted invalid block exceeding this limit by a small amount, I think last year, big oopsie by certain mining pool, lol). In times when mempool isn't full enough to fill a block with transactions equal and above fee rate of 1sat/vB it makes economical sense to include transactions that pay less than 1sat/vB when those are available in mempool. Not all mining pools do that, so far. Some do...

One upper limit for how many transactions can go into a block is 4,000,000WU (weight units, where 1 byte of witness data accounts for 1WU, while for non-witness transaction data 1 Byte accounts for 4WU. There's no relation between number of transactions in a candidate block and how long it takes to mine a valid blockheader hash. Depending on how many unconfirmed transactions are available in the mempool of a mining pool, a block candidate is commonly (quite simplified) assembled with transactions (or transaction packages in the case of child-pays-for-parents chains, parents must occur before child in block) from highest fee rate to lowest until the block is full (not exceeding 4M WU). This usually gives the mining pool maximum profit from available transaction fees in mempool. There are more limits to pay attention to, like number of Sigops can't exceed 80,000 (there has been at least one broadcasted invalid block exceeding this limit by a small amount, I think last year, big oopsie by certain mining pool, lol). In times when mempool isn't full enough to fill a block with transactions equal and above fee rate of 1sat/vB it makes economical sense to include transactions that pay less than 1sat/vB when those are available in mempool. Not all mining pools do that, so far. Some do...

One upper limit for how many transactions can go into a block is 4,000,000WU (weight units, where 1 byte of witness data accounts for 1WU, while for non-witness transaction data 1 Byte accounts for 4WU. There's no relation between number of transactions in a candidate block and how long it takes to mine a valid blockheader hash. Depending on how many unconfirmed transactions are available in the mempool of a mining pool, a block candidate is commonly (quite simplified) assembled with transactions (or transaction packages in the case of child-pays-for-parents chains, parents must occur before child in block) from highest fee rate to lowest until the block is full (not exceeding 4M WU). This usually gives the mining pool maximum profit from available transaction fees in mempool. There are more limits to pay attention to, like number of Sigops can't exceed 80,000 (there has been at least one broadcasted invalid block exceeding this limit by a small amount, I think last year, big oopsie by certain mining pool, lol). In times when mempool isn't full enough to fill a block with transactions equal and above fee rate of 1sat/vB it makes economical sense to include transactions that pay less than 1sat/vB when those are available in mempool. Not all mining pools do that, so far. Some do...