Optional, for the curious: explore what holding a faster research pace might require.
A calculator of what research would require under stated assumptions. It does not predict whether AI will supply that research, or whether an explosion happens.
Tejas Subramaniam, an AI strategy researcher, set out a short model of AI research.
If AI has doubled Anthropic's pace, Tejas's model says that's like having about 6 times the research workforce. On the same model, doubling the pace again would take about 30 times the original workforce.
Based on a short model by Tejas Subramaniam, using METR's estimate that Anthropic's models are improving about twice as fast as before on well-specified tasks.
The dials need JavaScript. Both views below show the default model.
Reference cases, using Tejas's starting assumptions. They do not change when you move the dials.
A step of progress takes
Pace
Researcher-equivalents
6 months
1×
1×
3 months
2×
about 6×
1.5 months
4×
about 30×
Compute beyond the old trend lowers these numbers. Harder ideas raise them. Test that with the dials below.
METR would not be surprised by 1.3× or 3×. Through this model that is about 2× to 16× the research labour. This is our calculation, not METR's.
The months are an illustration. METR measured a benchmark index, not release dates. These figures use Tejas's assumptions, and 4× is our what-if, not a forecast. These are starting figures. Holding the pace takes more each year if ideas get harder.
Some settings in the link were not valid, so their starting values are used.
What these inputs produce
With these inputs, AI progress runs at 2× its old pace.
What the target requires
To reach METR's 2× with everything else as set, research labour must be 6.25× its old level.
Changing research labour changes the first result, not the second.
With this setting, experiment computing alone produces some research even with no labour. That comes from the formula chosen. It is not evidence that research has become autonomous.
Old pace means METR's baseline, March 2024 to October 2025. We assume that trend would have continued. It is not an observed world without AI.
Change the inputsStarting inputs
The experiment dial holds training compute fixed. It changes the total compute pool and its split.
A simplified timing picture: experiment computing is used now, while training's share shows up over W years.
Researchers 6.25× → Research lab 2.5× → Stock of methods (software) → Training run → AI capability 2×.
Compute pool 1× splits into Experiments 1× → Research lab, and Training 1× → Training run.
METR would not be surprised by 1.3× or 3×
Methods are non-rival: each training run uses the whole stock. This assumes every training run can use every method at once, with no cost of adapting it. Older models stay available even if new runs get weaker. Pipe motion shows relative rates; the numbers give their values.
How the model works
Two things make AI models better: more training compute, and better methods, which we call software. Tejas assumes software supplies two-thirds of progress. If training compute keeps its pace, doubling overall progress needs software progress to run 2.5 times as fast.
Better methods come from a research lab. The lab needs researchers, human or AI, and computing power for experiments, and he weights them equally. Add researchers without adding experiment compute and each extra one adds less. That is why 2.5 times the pace of new methods takes 2.5 × 2.5 = 6.25 times the labour.
Research labour here means human researchers plus AI doing research work, measured in researcher-equivalents. An L of 6.25 means 6.25 times the old research labour, not 6.25 extra. It does not mean 6.25 times the staff. AI raises effective research labour when it adds research capacity or lets the existing team do more. Replacing a human researcher with equal AI capacity leaves this measure unchanged. Adding research labour adds no experiment compute.
Try these
Click a row to load it into the dials. The right-hand number is the research labour needed.
To double the pace (2×)
Tejas's modelHis assumptions, unchanged.6.25×
Research productivity falls 20%The same effort finds 20% less.9.77×
Twice the experiment computeTraining compute stays on its old path.3.13×
Twice the compute everywhereExperiments and training both double. Depends on our illustrative T₀ and W. METR is confident the acceleration was not due to faster training compute scaling.2×
AI caused 80% of the speedupMETR's guess. This is the labour equivalent of AI's part only, and it depends on where the other causes act.4.84×
Research labour and experiment compute are strong complementsResearchers, human or AI, share experiment compute, so extra labour helps less (σ = 0.75). This is about research labour against experiment computing. It does not capture bottlenecks between tasks inside research, such as research judgement against coding.9.41×
Research labour and experiment compute are very strong complementsPast a point, more researchers, human or AI, add almost nothing, and experiment compute sets a limit (σ = 0.5). This is about research labour against experiment computing. It does not capture bottlenecks between tasks inside research, such as research judgement against coding.Never reaches 2×: approaches a ceiling of 1.67×
To double the pace again (4×)
Tejas's modelHis assumptions, with the target at 4×.30.25×
Twice the experiment computeThe labour needed halves.15.13×
Four times the experiment computeThe labour needed falls to a quarter.7.56×
Research productivity falls 20%The same effort finds 20% less.47.27×
Where this comes from
Tejas called his post a scrappy summary, his loose impression of tweets by Ryan Greenblatt, Ajeya Cotra and others. He says falling research productivity could push the figure much higher. This page builds his model so you can change its assumptions. The starting software share, Cobb-Douglas form, labour weight, A and E are his, and the 2× is METR's. The compute split, the time path and the other settings are ours.
What the number depends on
These cautions are ours.
The model treats AI researchers as free of compute. In practice they run on compute too, which would compete with experiments and training.
Tejas supposes the two-thirds split and the equal weights. Change them and the answer moves.
METR's benchmark index gave 1.9×, with a 95% interval of 1.5× to 2.8×. Its 2× judgement also draws on other evidence, and it says Anthropic redacted most of what it found most informative.
Much of METR's evidence measures well-specified tasks, and METR calls its estimate uncertain.
The index and effective compute are different things. Reading one as the other assumes they move together.
METR guesses that perhaps 80% of the acceleration in October 2026 came from AI. Tejas's model gives AI all of it.
Tejas says falling research productivity could raise the labour figure much further. The 'Research productivity falls 20%' case shows one example.
What did the speedup cost? Anthropic's research spending rose about 5 times over the same year. We work through what that does and does not show on a separate page: The spending puzzle.
Inside the model
This is the same model as The machine, with its parts named. Two equations do the work: one makes new methods, the other makes better models. You can change anything here and the machine changes too. Hover over or tap any symbol to see what it means, or open In plain English under a card.
About this model
Two sectors, in growth rates. Time is in years; growth is in doublings per year. Every ratio compares with the assumed continuation of METR's baseline trend at the same date.
The research sector extends the Romer and Jones setup by adding experiment compute. The ideas-getting-harder extension draws on Bloom, Jones, Van Reenen and Webb, Are Ideas Getting Harder to Find? (American Economic Review, 2020). The compute split, the bridge and the time path are this page's own extensions. The paper does not calibrate θ here.
T0, W, θ, q₀, q₁ and L0, K0 defaults are illustrative. We did not estimate them.
A simplified timing picture: experiment computing is used now, while training's share shows up over W years.
L = 6.25, E = 1, A = 1 → g/g0 = 2.5 → S → C × S → AI capability 2×. K = 1, q0 = q1 = 0.5; c = 1 doubling a year.
METR would not be surprised by 1.3× or 3×
Follow the named parts into the equations below. The two views share every input.
Numbers in substitutions are rounded; ≈ marks approximate arithmetic.
Final sector · Training run
Effective compute = C × Sg0 = s × T0c0 = (1 − s) × T0
A model gets better in two ways: more training compute, or better methods. The two multiply, so doubling either one doubles effective compute. In growth rates the multiplying becomes adding: growth in effective compute equals growth in C plus growth in S. That lets us split baseline growth: s to software, 1 − s to training. Tejas supposes two-thirds and one-third. No research labour appears in this equation. Research labour comes in one step earlier, in the research lab. Methods are non-rival: each run uses all of S. This assumes every training run can use every method at once, with no cost of adapting it. Older models stay available even if new runs get weaker.
Symbols
C — Computing power used to train models
S — The stock of research methods
g0 — Baseline software growth in doublings/year
c0 — Baseline training growth in doublings/year
s — Software share of baseline progress
T0 — Baseline effective-compute growth in doublings/year
Research sector · Research lab
g = g0 × A × F(L, E)λG = g/g0 = A × F(L, E)λF(L, E) = La × E1 − a
With a = 0.5 and λ = 1, this is Tejas's equation: G = A × √L × √E.
This equation makes new methods. Its output is a pace: how fast S grows. Its inputs are research labour, human or AI, and experiment compute. With equal weights and λ = 1, neither input can do the work alone: four times the labour only doubles the pace. A scales the whole result. If ideas get harder to find, A falls and the same lab finds less. Cobb-Douglas builds in one fixed answer to how far more research labour can stand in for more experiments, so no symbol for it appears here. Choose CES below to set it yourself.
Symbols
g — Software growth in doublings a year
g0 — Software growth before the speedup
G — New software pace divided by old
A — Research productivity compared with baseline
L — Human and AI research labour
E — Experiment compute compared with baseline. Compute per unit of time
λ — Returns to scale in research
F — Combines labour and experiment compute
a — Labour weight against experiment compute
Compute link · Compute pool
E = k × q1/q0Cr = k × (1 − q1)/(1 − q0)c = c0 + log2(Cr)/W
A lab has one pool of compute and two uses for it: experiments and training. q₀ and q₁ are the shares that go to experiments before and after the change. If the pool stays the same, more for experiments means less for training. If the pool doubles and the split stays the same, both double. Experiments use a change at once. We assume training's effect on models arrives over W years. So E measures compute in use now, and Cr measures the training-compute effect, relative to baseline, at the end of W years. During those years the two need not add up to the pool. The experiment dial on The machine holds Cr fixed.
Symbols
k — Total compute pool divided by baseline
q0 — Experiment share before the change
q1 — Experiment share after the change
E — Experiment compute compared with baseline. Compute per unit of time
Cr — Training-compute effect relative to baseline at the end of W years
The model works in effective compute, but METR reports a benchmark index. This card turns one into the other. With h and b both at 1, the two speedups are the same. h says how much of the speedup AI caused. b says how the index speedup relates to the effective compute speedup. Only b × h enters. Lowering h means more of the reported speedup comes from outside AI, so the same inputs report a larger speedup. With h below 1, the labour figure covers only the part attributed to AI. The target X enters through Xe = 1 + b × h × (X − 1). Research productivity A enters through G in the research sector. For example, if other causes raised A to 1.3 and AI supplied the rest of a 2× pace, labour would need about 3.7×, against 4.84× with h = 0.8 and A = 1.
Symbols
G — New software pace divided by old
g0 — Software growth before the speedup
c — Training growth in doublings a year
T0 — Baseline effective-compute growth in doublings/year
Xe — Effective-compute speedup after h and b
Xout — The reported index speedup implied by the inputs, given b and h
b — Bridge from index to effective-compute speedup
At year 5: G1 ≈ 2.5G2 ≈ 2.5D(5) ≈ 15A(5) ≈ 0.0743L(5) = 1,130×.
In plain English
S is software, whose excess doublings enter through D(t). D(t) is how many more doublings S has had by year t than on the baseline path. It is negative if S has fallen behind. If D is negative, A rises above its starting ratio. If the other inputs beat the target with no labour, the lab keeps the faster growth. A change in training compute makes labour jump once, at year W: down after a cut, up after an increase. Read the shape of the line. Its size depends on θ and on g₀, which comes from T₀. Both are our illustrative numbers.
Symbols
A(t) — Research productivity at year t
A — Research productivity at the starting point
θ — Rate at which ideas get harder
D(t) — How many more doublings S has had by year t than on the baseline path. It is negative if S has fallen behind.
G1 — Required software speedup through W years
G2 — Required software speedup after W years
Xe — Effective-compute speedup required for the target
T0 — Baseline effective-compute growth in doublings/year
g0 — Software growth before the speedup
c — Training growth through W years
c0 — Baseline training growth after W years
W — Years for the training compute change
t — Time since the change in years
max — The larger of two values
Grequired — Software growth required for the target
Gactual — Software growth the lab actually produces
Gfloor — Software growth at zero labour
F — Combines labour and experiment compute
E — Experiment compute compared with baseline. Compute per unit of time
λ — Returns to scale in research
Set the assumptions
From the target to labour
Target X ≈ 2 → Xe = 1 + 1 × 1 × (2 − 1) ≈ 2.Why this step
Start from the target speedup. Adjust it for how much AI caused and for how the index relates to effective compute.
G = (Xe × T0 − c)/g0 = (2 × 3 − 1)/2 ≈ 2.5.Why this step
Take away what training compute supplies. Software must supply the rest.
Tg = (G/A)1/λ = (2.5/1)1/1 ≈ 2.5. Tg is the required value of F.Why this step
Divide by research productivity, then undo the power λ. This gives what the lab's research labour and experiment compute must produce.
L = (Tg/E1 − a)1/a = (2.5/11 − 0.5)1/0.5. L ≈ 6.25×.Why this step
Undo the production function to find the labour that produces it.
Progress as labour rises
At 6.25× labour, progress runs at 2×. Labour uses a logarithmic axis.
Labour ratio on a log scale. METR would not be surprised by 1.3× or 3×. The dotted horizontal line marks 2×. The dot shows your forward input.
Labour to hold the pace
These figures depend on our illustrative T₀. At current settings, with T₀ = 1: L(5) = 35.4×. With T₀ = 6: L(5) = 205,000×.
Target 2×. At year 0: 6.25×. L(5) = 1,130×. Labour uses a logarithmic axis. Muted dashes hold θ = 0.
Years from now; labour ratio on a log scale. Zero labour is stated separately. The dashed reference holds θ at zero. A wall ends the line. Very large labour ratios use number words. This is a sensitivity exercise. We did not estimate these parameters for Anthropic, and the chart does not predict labour supply or model AI feeding back into research labour.
Worked example: holding the pace
Suppose AI doubles the pace. On the starting model, that is like having about 6.25 times the research workforce, counting human researchers and AI research work together. To hold that pace for five years, the workforce would need to be about 1,130 times the workforce on the old path at that date.
The yearly growth factor is how many times more research effort each year keeps the old pace. On the starting model, the settings for research difficulty and returns to effort set it. The calculation multiplies 6.25 by the yearly growth factor raised to the 15th power: three powers for each of five years. The three comes from the starting settings for how much faster software must improve. The starting factor is 1.414, the square root of two; its 15th power is about 181. Using Anthropic’s slower spending growth in 2025, about 2.8 times a year on the lowest count from the spending puzzle, as a stand-in gives a factor of 2.786, whose 15th power is about 4.7 million, giving about 30 million times the workforce on the old path. A small change in this uncertain factor makes a large difference.
These are extreme illustrations, not forecasts. They assume the doubled pace is held, training follows its set path, labour and computing combine as in the starting model, experiment computing stays a fixed share, and research gets harder at a steady rate. The second case also treats spending growth as research-effort growth, which the spending puzzle says is uncertain. More experiment computing, a change in research, or a slower pace could change the result.