Chart data could not load; reload to try again.
Artificial Intelligence
Capability gained at equal spending
Chained ECI gains at matched token budgets, relative to March 2023.
Methodology
Assumptions
This chart compares Epoch AI's ECI scores and input and output token prices from the November 9, 2026 snapshot.[^1] It uses 195 distinct model versions with an ECI score, release date, and both token prices. Duplicate model/date/score/price rows are counted once. Different dated or priced versions remain eligible.
Prices are held at the snapshot value for every release date, including retired endpoints. This is a release-date comparison using current prices, not a history of prices paid at the time. The basket contains 750,000 input tokens and 250,000 output tokens; it standardizes prices rather than estimating the tokens a task needs.
Formulas and methods
For a model, basket cost is 0.75 times the input price plus 0.25 times the output price, in dollars per million tokens. At a given date and budget, the frontier is the highest ECI among models released by then whose basket costs no more than that budget.
For each release date, compare the frontiers before and after adding every model released that day. The lower budget L is the higher of the cheapest available basket costs before and after the release. The upper budget U is $93.75, the largest eligible basket price in the snapshot. Integrate the change across that common affordable range with equal weight for equal proportional changes in price, then divide by the width of the range in log dollars:
basket_cost = 0.75 × input_price + 0.25 × output_price
frontier(date, budget) = max ECI among eligible released models
release_gain = integral[L,U] (frontier_after(B) - frontier_before(B)) d ln(B)
/ ln(U/L)
chained_gain = sum of release_gain since March 14, 2023If L equals U, use the frontier difference at that budget. The lower boundary can fall as cheaper models arrive; a newly affordable budget enters only when a previous frontier exists there.
Walkthrough
Save Epoch's model table and calculate each eligible basket. GPT-4's March 2023 version costs 0.75 × $30 + 0.25 × $60 = $37.50. Set the index to zero at its release. For each later release date, calculate the before and after price frontiers, integrate their difference over their shared budget range, and add the resulting gain to the running total. The chart shows that cumulative ECI gain, not an absolute ECI level.
Notes
The budget ceiling, changing lower boundary, and log-price weighting affect the result. The chain combines comparisons over different budget ranges; it is not the gain at a single fixed dollar budget. Missing scores or prices, pricing changes, provider differences, caching, batch discounts, tool fees, and variation in token use can change the comparison. The chart does not convert the gain to task time or show uncertainty intervals.
The downloadable package contains the frozen source, derived priced rows, and script. Run python3 script.py to reproduce the chart data with Python 3.11 or newer; the upstream ECI fitting process itself is not reproduced.
Rolling cost of AI Model Capabilities
The cheapest token basket at each ECI score, grouped by model release date.
Methodology
Assumptions
The chart uses Epoch AI's ECI scores and input and output token prices from the November 9, 2026 snapshot.[^1] It includes 195 distinct model versions with a score, release date, and both prices. Duplicate model/date/score/price rows are counted once. Different dated or priced versions remain eligible.
A token basket contains 750,000 input tokens and 250,000 output tokens. Prices are held at snapshot values across release dates, including retired endpoints; the colors do not reconstruct historical prices. The basket standardizes prices rather than estimating the tokens a task needs.
Formulas and methods
basket_cost = 0.75 × input_price + 0.25 × output_price
cheapest(date, ECI) = min basket_cost among released models with score ≥ ECIFor each release date and each distinct eligible ECI threshold, select the cheapest qualifying model available by that date. The upper edge is the best available ECI, and blank space means no priced model qualifies. The display spans ECI 110–170 and uses no annual bins. Colors use a logarithmic price scale from $0.03 to $100; values outside that range use the nearest endpoint color.
The hover detail also gives a modeled METR TH1.1 80% task horizon, not a measured horizon for that ECI level. Fit the log of METR p80 hours to ECI for 20 matched models using the tracker's version matches.[^2][^3]
ln(p80_hours) = -22.20441287 + 0.14432844 × ECI
n = 20; R² = 0.970768Walkthrough
Calculate the basket cost for every eligible row. At a selected date and target, such as ECI 150, discard models not yet released and those below ECI 150, then take the lowest remaining basket cost. Repeat at every release date and threshold. The hover's optional time estimate at ECI 150 is exp(-22.20441287 + 0.14432844 × 150) = 0.574 hours, about 34 minutes.
Notes
Missing prices and scores are excluded. Real costs may differ with provider, context length, cache or batch discounts, tool fees, and tokens used. The displayed ECI and cheapest price are based on the snapshot; their accuracy depends on the source scores and listed prices.
The hover's ECI-to-time fit has residual standard deviation 0.3128 log hours, roughly a 1.37× scatter factor. The matched scores span ECI 121.87–157.82, so hover estimates outside that range extrapolate. Model matching, benchmark scores, scaffolds, and METR sampling add uncertainty; the scatter is not a full prediction interval. METR cautions that horizons above 16 hours are unreliable with its current task suite.[^2]
The downloadable package contains the frozen source, priced rows, calibration pairs, and script. Run python3 script.py to reproduce the chart data with Python 3.11 or newer; the upstream ECI and METR experiments are not reproduced.
Task-horizon progress
Record-setting task horizons and three trajectories, including modeled estimates.
Methodology
Assumptions
Task horizon is the human expert duration of a software task that an agent is predicted to complete with 80% success on METR's TH1.1 suite.[^2] It is not AI runtime or a claim about all kinds of work. The chart combines direct measurements with modeled estimates; converted TH1.0 values and ECI- and AECI-based values are estimates on the TH1.1 scale.
For ECI-based estimates, fit log TH1.1 p80 hours to ECI using 20 matched model versions from the saved tracker and Epoch data.[^1][^3] The source matches and exact inputs are in calibration.csv.
ln(p80_hours) = -22.20441287 + 0.14432844 × ECI
n = 20; R² = 0.970768; residual SD = 0.3128 log hoursThe ECI calibrators span 121.87–157.82. Estimates outside that range extrapolate. Epoch includes METR-derived benchmark evidence, so the correlation is not an independent validation of METR. Differences in scaffold, benchmark, and token budget add uncertainty. METR says horizons above 16 hours are unreliable with its current task suite.[^2]
Coverage
The task-horizon input contains all 724 scored Epoch source rows, represented by 259 distinct model/date/score entries, plus 23 METR TH1.1 p80 measurements, 31 TH1.0 p80 measurements, and 19 Anthropic AECI entries.[^1][^2][^3][^6] The chart selects one value for each matched model and date in this order: published TH1.1 p80, converted TH1.0 p80, ECI-based estimate, then AECI-based estimate. When the same source has more than one score for a model and date, the highest horizon is selected. These decisions leave 270 selected model/date entries from 332 inputs; all source entries remain in all-horizons.csv.
Most matches use the same model name and date. Explicit source-backed aliases reconcile GPT-4o's May 2024 version, Claude 3 Opus, o1, and GPT-4 Turbo Preview's January 2024 version when source labels or dates differ. For Claude 3 Opus and o1, the selected measurement date is used. The tracker dates GPT-3.5 Turbo Instruct to March 15, 2022; the chart instead uses September 18, 2023, the version release date in Epoch's model table.[^1][^3] Three entries in METR's TH1.1 file are labeled TH1.0. They are converted as TH1.0 rather than counted as independent TH1.1 measurements.[^2][^6] Other distinct dated versions are retained. The raw CSV retains all inputs and estimates.
The chart starts on January 1, 2023. By default, a dot is shown only when its horizon exceeds every model available on an earlier date, including before 2023. If models released on the same date tie for the new maximum, each is included. This gives 21 frontier dots on 21 dates. Show all models displays 268 selected entries: all dated entries from 2023 onward, plus four with no release date in the source: Claude Instant 1.1, Gemma 2 9B, GPT-5 Chat, and Gemma 2 27B. They appear in an Undated column only when all models are shown; they are excluded from the frontier and trend fits. No date is invented. All dots use the same visual style; the source of each value remains in the model data and this methodology.
The METR report also names an Unreleased 2022 model, but the saved AECI extraction provides no score and dated marker for it. It is not assigned an invented estimate.[^4]
Converting TH1.0 to TH1.1
Match the 16 named models measured on both TH1.0 and TH1.1. The explicit ID mapping is in script.py; the source measurements are in metr-results-1-0.yaml and metr-results.yaml.[^6][^2] Fit ordinary least squares with the natural log of TH1.0 p80 minutes as the input and the natural log of TH1.1 p80 hours as the output. The three TH1.0 results carried into the TH1.1 file are excluded. With H₀ in minutes and H₁ in hours:
ln(H₁) = -5.11084335 + 1.29457173 × ln(H₀)
n = 16; R² = 0.974065; residual SD = 0.272182 log hours
GPT-4 0125: H₀ = 1.913986 minutes → H₁ = 0.8386 minutesThe conversion is an estimate on the TH1.1 scale, not a TH1.1 measurement. METR's suites and agent scaffolds differ, and uncertainty in both versions is omitted. Three early models use converted TH1.0 values rather than treating a copied TH1.0 value as TH1.1. The fit's TH1.0 inputs span 1.37–40.04 minutes; conversions outside that range are extrapolations. The ECI score for GPT-4 0125 is shared by Epoch's broader model group and is not a separate measurement for that version.[^1]
Holdout comparison
For each model in a comparison, remove that model from every method's calibration, refit, then predict its measured TH1.1 p80. Score errors in log hours and report exp(RMSE): 1 is perfect and lower is better. This is model holdout, not a historical forecast. Each method first trains on all its available matched models except the held-out model; the models being scored are identical within each row. holdout_comparison.py reproduces the results.
Models with both TH1.0 and ECI: n = 15 TH1.0 1.347 ECI 1.376
Models with TH1.0, ECI, and AECI: n = 6 TH1.0 1.377 ECI 1.419 AECI 1.263When training is restricted to the same shared models too, TH1.0 and ECI score 1.351 and 1.354 on the 15-model comparison. TH1.0 wins 7 of those model holdouts and ECI wins 8. The TH1.0 advantage is small and not conclusive; on the six Anthropic models AECI scores best. The source order is a display policy, not a demonstrated accuracy ranking for every model. The held-out outcomes are excluded from fitting, but the published ECI and AECI inputs were built using full upstream benchmark snapshots, so they are not independent historical forecasts. Shared task suites, scaffolds, digitization, uncertain version matching, small samples, and benchmark saturation limit the comparison. In particular, the six-model AECI comparison does not validate extrapolation to internal frontier models.
Anthropic's index: assumptions and fit
AECI scores come from the same October fit shown in METR's November 5 assessment, using the tracker's extracted values and reported anchors.[^3][^4] The raw figure is included as aeci-main.png with a checked hash. Most scores are digitized to about 0.1 AECI; Opus 5.5 and internal Model 4 have reported anchors. Public dates are release dates. Internal dates are approximate dates read from the figure for availability to most Anthropic employees, not public releases or precise training completion dates.
Fit the log of exact METR p80 hours to AECI for all eight entries with an exact METR model ID, including the early Mythos Preview snapshot. The public Mythos Preview snapshot is a distinct entry. AECI is fitted directly to METR rather than treated as an Epoch score: an additional AECI-to-ECI bridge would add another model assumption, and the tracker reference uses the direct fit.[^3]
ln(p80_hours) = -22.68266456 + 0.1470880061 × AECI
n = 8; R² = 0.989430; residual SD = 0.221025 log hours
Model 2: AECI 174.04 → 18.4771 hours
Model 3: AECI 176.98 → 28.4734 hours
Model 4: AECI 178.60 → 36.1347 hoursThe AECI fit is reproduced from the saved scores and full-precision METR values, then applied to all 19 AECI entries before model selection. These are modeled horizons, not METR task measurements. The eight calibrators cover AECI 124.71–162.35; Models 2–4 are beyond that range. Their results rely on continued linearity in log time. The chart shows central estimates; the fit scatter is not a complete uncertainty range. Digitization, index error, vendor benchmark selection, saturation, small sample size, and METR's measurement limits add uncertainty.
Frontier trend lines
The single exponential and two-breakpoint lines are fitted to the same 21 frontier dates, from February 24, 2023 through September 25, 2026. Each date has equal weight. With H as task horizon in hours and t as days since January 1, 2025 divided by 365.2425, ordinary least squares minimizes squared errors in ln(H). The single exponential is:
ln(H) = -1.79770097 + 2.63350452 × tFor the two-breakpoint line, choose two dates that minimize the same squared error, with at least three frontier dates before, between, and after the breakpoints. Search a 14-day grid, then search daily within 14 days of the best grid pair. This deterministic search is an approximation, not a guarantee of the global daily optimum. The fitted breakpoints are April 10, 2024 and February 16, 2026:
ln(H) = -2.64045585 + 1.77457981 × t
+ 0.81277178 × max(0, t + 0.72828326)
+ 1.75218850 × max(0, t - 1.12527978)The lines are continuous at both breakpoints and become dashed after the last frontier date. The AI 2027 line is the separate scenario described below. The fitted lines extend to April 27, 2027; this extension is a projection, not a measured result. The curves depend on the chosen snapshot, model matching, priority rule, and record selection. The late frontier includes estimates above the range where METR says its task suite can measure reliably.[^2] The two-breakpoint line has more flexibility and is fitted to the same data used to choose its breakpoints, so its close fit to those points is not evidence that it will predict future results better. No confidence band is calculated here.
AI 2027 comparison
The dashed red line is a scenario, not a measured series or a new forecast. It uses the reconstruction in the saved tracker source, checked there against the authors' grading sheet.[^3][^5] Let u be the UTC decimal year minus 2025. Evaluate:
ln(seconds) = 6.36588017 + 1.75547434 u + 0.350855496 u²
+ 0.0462100721 u³ + 0.0144760767 u⁴
+ 0.0194798378 u⁵ + 0.00743193880 u⁶
+ 0.000851343348 u⁷
raw_hours = exp(ln(seconds)) / 3600
TH1.1_hours = raw_hours × rebase
rebase = Claude 3.7 Sonnet's measured TH1.1 p80 hours
/ raw_hours on April 3, 2025
≈ 0.783This anchors the older TH1.0 scenario scale to the best model at the scenario's publication date. A constant rebase is an approximation: the two task suites do not have a uniform ratio across models. The polynomial after the authors' graded window is a mathematical continuation. A model's position relative to the curve is therefore a comparison of estimates with a scenario, not evidence that its real-world capabilities are ahead of a forecast. The vertical dashed line marks the November 9 data snapshot.
Reproducing the complete model chart
Run python3 script.py in the extracted package. It emits the complete chart JSON, regenerates the priced data.csv and all-horizons.csv, selects the plotted points and frontier, fits the calibration and frontier trend lines, and reconstructs the scenario. Run python3 holdout_comparison.py to reproduce the holdout results. all-horizons.csv contains all 332 source entries with their measurement/estimate series and source; the JSON contains the 270 selected entries and 21 frontier points from 2023 onward. No unavailable price is filled and no model without a published score is given one. The source figures and frozen tracker files preserve the upstream extraction choices, but do not recreate the underlying private Anthropic benchmark experiments.