Many factors need to be taken into consideration in order to increase the effective strength of a humanlike form in a realistic manner. One can always scale up just a few of these factors and say, "Wow, look at how strong this would be!", but when put into practise in a real-world situation, something would either fail or limit the naïvely-calculated maximum exertion. This is a problem of engineering, not simply physics and materials science, and trade-offs will always have to be made.
Strength of the materials: Stronger materials can withstand more force before being damaged. Any mismatch between actuator and skeletal strength increases the likelihood of mechanical failure.
Density of the materials: The lighter the materials, the less force will be required to move objects made from them.
The maximum force that can be applied by the actuators per unit mass: The more force that can be applied per unit mass, the less will be needed to simply move the individual's limbs, and the more that can be applied to the environment.
The internal friction of the actuators and joints. The higher, the less efficient - and slower - the humanoid will be.
The maximum speed of the actuators. Additional speed can be achieved by chaining actuators in series, but increasing strength requires that they be placed in parallel.
The lever ratio of the humanoid's joints: Higher strength means lower speed.
The maximum allowable time for a joint to transition between fully flexed and fully extended: The humanoid might otherwise be immensely strong, yet unable to outpace a snail.
The energy efficiency of the actuators. Less efficient actuators require more energy input - and generate more waste heat, limiting the maximum duration of maximum exertion, as well as limiting the maximum number of actuators for the available cooling capacity and thermal tolerance. Given that no mechanism is ever 100% efficient, by rights, if his muscles were just as efficient as ours, Superman ought to glow visibly from the waste heat when he applies his super strength, though perhaps that's where the heat for his eyes' heat rays comes from...
The thermal tolerance of the humanoid: Raise a human's temperature more than a few degrees, and the efficiency and efficacy of all sorts of things is significantly reduced. The same applies to most physical systems outside a vacuum.
The cooling capacity of the humanoid: Humans have amongst the best cooling capacity of any species on earth, allowing a sufficiently fit individual to literally run down members of most other terrestrial animal species solely on the basis that a human can maintain an optimum body temperature while running at a pace at which causes pretty much every other animal species' body temperature to increase. Keep this up long enough, and the prey's rising body temperature will eventually cause a physical collapse. Additionally, with insufficient internal heat redistribution capacity, local temperature differences could increase to the point where localised damage would occur.
The humanoid's available energy and its maximum rate of delivery to the actuators: There's no point in having actuators that consume vast amounts of power if sufficient power cannot be supplied to them, or if sufficient power can be supplied, but not for a sufficient amount of time.
Whether the humanoid's actuators require an input of energy only when changing position or if energy is required to maintain a given force - i.e. electric screw-jack vs a long-travel electromagnetic linear actuator; the former requires power only to change its position, while the latter requires (more) power to maintain a given position somewhere between fully extended and fully contracted, and less (or none) when at the limits of its travel: Animal muscles fall into the latter category, increasing the basal metabolic cost of things as simple as maintaining a static standing posture, but if a humanoid had actuators that fell into the former category, only actuators actively moving would consume power, though maintaining a standing posture can be a rather more active task than it might seem.
The humanoid's optimum operating temperature: If this is too different from the environmental temperature, then energy must be expended to maintain - or achieve - it, or the humanoid will suffer from losses of efficiency or from mechanical failure if its internal cooling or heating capacity cannot keep up with the rate of heat transfer, which increases proportionally to the difference in temperature.
The humanoid's skin's insulation factor. By reducing the speed at which heat is gained or lost, this can increase the efficiency of the humanoid at rest, but can hamper the task of dealing with waste heat.
The environmental temperature: While the average temperature is important when optimising a real mechanical system, the range of variation is also important. The lower the range of variation, the easier it is to optimise a system to operate in those conditions. Being able to tolerate a wider variety of conditions can be useful in being able to survive, but also increases the cost of insulation and cooling, and systems that accept a narrower range of acceptable conditions have lower metabolic requirements than those that accommodate all possible conditions, though at the cost of an increased probability of exposure-related injury or death.
The mass of any support equipment required to allow the humanoid to function independently for a reasonable amount of time. Unless this humanoid is like an evangelion with a cable to its power supply - which would have its own problems - it'll have to carry everything it needs around with it. If it needs a lot of energy, it better have a compact way of storing that energy.
So, as can be seen, this is a very difficult problem. Sure, you may want a 'superman', but you have to answer all of these questions - and more - before you can even begin to enumerate his realistic capabilities.
However, let's assume that we need a 'superman' with roughly human capacities for duration of exertion and with equivalent environmental optimums. We would need higher-strength materials, more powerful, lighter and more efficient actuators, and probably better cooling capacity too. Given intelligent design rather than evolution, we could achieve results that could never evolve naturally. We are still limited by real-world chemistry and physics, though.
It is not realistic to give a period of maximum exertion of as little as 10 seconds unless in that time your humanoid can achieve everything that a human could achieve in perhaps as much as 5 minutes. If your humanoid was this fast, it would have to have traded off a lot of strength - both structural and the force it could apply to external objects - to achieve that speed.
For a superhero, a realistic duration of elevated exertion would be more like 5 to 10 minutes. As a martial artist, practising karate and hapkido, at my dojo, the examination for black belt includes engaging in ten continuous minutes of fighting. Since a superhero might be able to achieve his results a bit faster than a mere human, five minutes might be a reasonable compromise.
If we build our superhero with light, strong materials like carbon fiber (perhaps 20 times stronger than human bone), use super-strong manufactured muscles that are up to a hundred times stronger than a human muscle like these, and provide a LENR atomic power supply, then it is conceivable that a human-sized and shaped form could lift (in an event such as the clean and jerk in Olympic weightlifting) not 263kg (the current human world record), but something on the order of 10,000kg.
However, being able to lift large weights is not all there is to being a superhero. Being able to lift ten tons at the same speed as a human doesn't mean that you punch any harder unless the actuators are also faster - if your arm weighs as much as a human's and accelerate just as fast, then the impact force will be the same. However, such a superhero could simply pick up a 50kg hand barbell in each hand and still get his punches out just as fast. Given the physics formula $e= 1/2mv^2$, artificially increasing the mass of a perhaps 10kg arm by 50kg and still hitting at the same impact speed would increase the impact energy by a factor of six, changing a punch from something that might break a bone if carefully placed to something far more likely to break bones every time it landed, or fatally concuss a human with a single punch most of the time.
However, if our intelligently-designed superhero was to do this for any length of time, say in a five-minute bout of all-out combat, then he would be generating a great deal more waste heat than any human in the same situation, requiring far more cooling capacity than even a 'merely' evolved human body can muster. His breath might be like a hair-dryer and/or his skin might literally steam with the amount of waste heat that would have to be dealt with.
On the other hand, if we gave our artificial superhero strength not a great deal greater than a human's - perhaps twice as strong at most - but a much higher speed, then he might be delivering punches that land at not ~9 m/s (32 kph or 20 mph), but at ~63 m/s (227 kph or 140 mph). That's 7 times the speed, but, because $e=1/2mv^2$, means that the punches will deliver 49 times the energy. That's the difference between bruising and perhaps a broken bone from a human fighter to a punch from our superhero that could almost literally knock the other guy's head off.
In order to control a body this responsive, we'd need to give our superhero light-speed electrical signalling instead of the downright sluggish electrochemical system our own nerves employ. This would also mean that our artificial superhero could literally watch a human opponent throw a punch at him, then - before that punch hit - throw six punches in return, each one potentially fatal or crippling before finally blocking the incoming punch, assuming that energy transfer to his opponent didn't knock him so far back as to make blocking the initial punch entirely unnecessary.
