Surse și atribuire

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'arc_agi_2_external.csv' table and rescale its 0-1 scores to percent. Underlying results published by ARC Prize — https://arcprize.org/leaderboard, transcribed by Epoch AI.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'agent' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'document_style_control' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'text_factuality' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'image_edit' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'search_style_control' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'text_style_control' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'text_to_image' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'text_to_video' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'vision_style_control' config.

Source: LMArena / arena.ai Leaderboard Dataset (lmarena-ai/leaderboard-dataset) on Hugging Face — https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the latest snapshot of the 'webdev' config.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Berkeley Function Calling Leaderboard (Gorilla, UC Berkeley) — https://gorilla.cs.berkeley.edu/leaderboard.html, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'bfcl' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); hardware runs of the metriq-gym "BSEQ" benchmark (Apache 2.0) as published to Metriq (https://metriq.info). Modified by aiqbench: one row per device, the most recent eligible run, score = the largest number of qubits whose Bell-state effective-qubit (BSEQ) test passed. Runs on simulators are not ranked.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); hardware runs of the metriq-gym "CLOPS" benchmark (Apache 2.0) as published to Metriq (https://metriq.info). Modified by aiqbench: one row per device, the most recent eligible run, score = CLOPS as metriq-gym measures it (100 qubits, 100 layers, twirled). Only runs with num_qubits = 100, num_layers = 100 are ranked. Runs on simulators are not ranked.

Source: Awesome Quantum Computing Experiments, François-Marie Le Régent, https://github.com/francois-marie/awesome-quantum-computing-experiments (data/entangled_state_error_exp.csv), MIT License (https://opensource.org/license/mit). Cite as: Le Régent, F.-M. (2025), "Awesome Quantum Computing Experiments: Benchmarking Experimental Progress Towards Fault-Tolerant Quantum Computation", https://arxiv.org/abs/2507.03678. Modified by aiqbench: one row per CSV record, ranked on the reported entangled-state or two-qubit gate error (lower is better); every column is kept in metadataJson.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); hardware runs of the metriq-gym "EPLG" benchmark (Apache 2.0) as published to Metriq (https://metriq.info). Modified by aiqbench: one row per device, the most recent eligible run, score = error per layered gate (EPLG) at the 100-qubit chain when the run measured it, otherwise at the longest chain length with a value; the chain length is recorded with every row and lower is better. Runs on simulators are not ranked.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'frontiermath.csv' table and rescale its 0-1 scores to percent. Evaluation run by Epoch AI.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Holistic Agent Leaderboard (Princeton) — https://hal.cs.princeton.edu/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'hal-gaia' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'gpqa_diamond.csv' table and rescale its 0-1 scores to percent. Evaluation run by Epoch AI.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'hle_external.csv' table and rescale its 0-1 scores to percent. Underlying results published by Scale AI / CAIS — https://agi.safe.ai/, transcribed by Epoch AI.

Source: IBM Quantum Platform API (https://quantum.cloud.ibm.com/api/v1/backends) — IBM's own published system facts for each quantum computer the instance can reach (physical qubit count, processor family, CLOPS, queue length) and its latest calibration (Qiskit BackendProperties), read through the vendor's REST API with an IBM Cloud account. The median two-qubit gate error, T1, T2 and readout error are medians over IBM's calibration data as published; not an aiqbench measurement. Score = physical qubits. Subject to IBM Quantum's terms of use.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Artificial Analysis — https://artificialanalysis.ai/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'artificial-analysis-llms' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent. Attribution to artificialanalysis.ai is required by their terms.

Source: IonQ Quantum Cloud API (https://api.ionq.co/v0.4/backends) — IonQ's own published system facts for each QPU (status, physical qubit count, queue time, native gate set, error-mitigation options), read through the public backends endpoint; with an API key, the vendor's characterization figures (fidelity, timing) as reported. Not an aiqbench measurement: the numbers are the vendor's, unmodified. Score = physical qubits. See https://docs.ionq.com for the API's terms.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); hardware runs of the metriq-gym "Linear Ramp QAOA" benchmark (Apache 2.0) as published to Metriq (https://metriq.info). Modified by aiqbench: one row per device, the most recent eligible run, score = effective approximation ratio (metriq-gym's score: the approximation ratio rescaled against random sampling, 0 when not above it) on a 1-D graph at 20 qubits with p = 10 layers. Only runs with graph_type = "1D", num_qubits = 20, qaoa_layers = [10] are ranked. Runs on simulators are not ranked.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by LiveBench — https://livebench.ai/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'livebench' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Artificial Analysis — https://artificialanalysis.ai/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'artificial-analysis-llms' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent. Attribution to artificialanalysis.ai is required by their terms.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); hardware runs of the metriq-gym "Mirror Circuits" benchmark (Apache 2.0) as published to Metriq (https://metriq.info). Modified by aiqbench: one row per device, the most recent eligible run, score = the polarization (success probability rescaled so random guessing scores 0) of 16-qubit, 32-layer randomised mirror circuits. Only runs with width = 16, num_layers = 32 are ranked. Runs on simulators are not ranked.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Artificial Analysis — https://artificialanalysis.ai/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'artificial-analysis-llms' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent. Attribution to artificialanalysis.ai is required by their terms.

Source: MMLU-Pro Leaderboard (TIGER-Lab, University of Waterloo) — https://huggingface.co/datasets/TIGER-Lab/mmlu_pro_leaderboard_submission. Licensed under Apache 2.0. Modified by aiqbench: we ingest the 'results.csv' file from the HF dataset root, rescale each row's Overall accuracy from 0-1 to percent, and preserve every disciplinary sub-score (Biology, Business, Chemistry, Computer Science, Economics, Engineering, Health, History, Law, Math, Philosophy, Physics, Psychology, Other) plus the Model Size and Data Source (TIGER-Lab vs Self-Reported) verbatim in each entry's metadata. Rows self-reported by model authors are labelled '(self-reported)' in the display name so readers can see the provenance without opening the details panel.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'otis_mock_aime_2024_2025.csv' table and rescale its 0-1 scores to percent. Evaluation run by Epoch AI.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Artificial Analysis — https://artificialanalysis.ai/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'artificial-analysis-llms' collection's headline result per model, normalised into aiqbench's common board format. Median output tokens per second on the provider's API; attribution to artificialanalysis.ai is required by their terms.

Source: Awesome Quantum Computing Experiments, François-Marie Le Régent, https://github.com/francois-marie/awesome-quantum-computing-experiments (data/physical_qubits.csv), MIT License (https://opensource.org/license/mit). Cite as: Le Régent, F.-M. (2025), "Awesome Quantum Computing Experiments: Benchmarking Experimental Progress Towards Fault-Tolerant Quantum Computation", https://arxiv.org/abs/2507.03678. Modified by aiqbench: one row per CSV record, ranked on the energy-relaxation time T1, converted from seconds to microseconds (T2 kept alongside); every column is kept in metadataJson.

Source: Awesome Quantum Computing Experiments, François-Marie Le Régent, https://github.com/francois-marie/awesome-quantum-computing-experiments (data/qubit_count.csv), MIT License (https://opensource.org/license/mit). Cite as: Le Régent, F.-M. (2025), "Awesome Quantum Computing Experiments: Benchmarking Experimental Progress Towards Fault-Tolerant Quantum Computation", https://arxiv.org/abs/2507.03678. Modified by aiqbench: one row per CSV record, ranked on the number of physical qubits controlled or entangled; every column is kept in metadataJson.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0; hardware runs of the QED-C Quantum Fourier Transform circuit executed with metriq-gym (Apache 2.0), which wraps the QED-C application-oriented benchmark harness (SRI-International/QC-App-Oriented-Benchmarks, Apache 2.0). Modified by aiqbench: one row per device, the most recent run that matches one protocol — QFT swept over 4, 8 and 12 qubits, 3 circuits per width, 1000 shots, method 1, no error mitigation. Score = metriq-gym accuracy_score (mean circuit fidelity across the sweep, 0–1). Where a vendor uploaded three single-width runs on one day instead of a sweep, the same mean and pooled uncertainty are computed from them and the row is marked derived. Runs with other widths or parameters are not ranked.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); hardware runs of the metriq-gym "QML Kernel" benchmark (Apache 2.0) as published to Metriq (https://metriq.info). Modified by aiqbench: one row per device, the most recent eligible run, score = quantum-kernel classification accuracy (0–1) at 20 qubits, 1000 shots. Only runs with num_qubits = 20 are ranked. Runs on simulators are not ranked.

Source: Quantum Advantage Tracker (quantum-advantage-tracker/quantum-advantage-tracker.github.io) — https://quantum-advantage-tracker.github.io/. Licensed under Apache 2.0. Modified by aiqbench: we ingest the 'CVP' pathway's submissions.json verbatim, one LeaderboardEntry per submission. Cross-instance comparisons are NOT meaningful per QAT's editorial policy — the panel groups by instance and refuses to rank across instances. Rich per-submission fields (runtime, compute resources) are not stored; each entry links back to its GitHub issue for full provenance.

Source: Quantum Advantage Tracker (quantum-advantage-tracker/quantum-advantage-tracker.github.io) — https://quantum-advantage-tracker.github.io/. Licensed under Apache 2.0. Modified by aiqbench: we ingest the 'OE' pathway's submissions.json verbatim, one LeaderboardEntry per submission. Cross-instance comparisons are NOT meaningful per QAT's editorial policy — the panel groups by instance and refuses to rank across instances. Rich per-submission fields (runtime, compute resources) are not stored; each entry links back to its GitHub issue for full provenance.

Source: Quantum Advantage Tracker (quantum-advantage-tracker/quantum-advantage-tracker.github.io) — https://quantum-advantage-tracker.github.io/. Licensed under Apache 2.0. Modified by aiqbench: we ingest the 'VP' pathway's submissions.json verbatim, one LeaderboardEntry per submission. Cross-instance comparisons are NOT meaningful per QAT's editorial policy — the panel groups by instance and refuses to rank across instances. Rich per-submission fields (runtime, compute resources) are not stored; each entry links back to its GitHub issue for full provenance.

Source: Awesome Quantum Computing Experiments, François-Marie Le Régent, https://github.com/francois-marie/awesome-quantum-computing-experiments (data/qec_exp.csv), MIT License (https://opensource.org/license/mit). Cite as: Le Régent, F.-M. (2025), "Awesome Quantum Computing Experiments: Benchmarking Experimental Progress Towards Fault-Tolerant Quantum Computation", https://arxiv.org/abs/2507.03678. Modified by aiqbench: one row per CSV record, ranked on the largest code distance d demonstrated, parsed from the [[n,k,d]] code parameters; every column is kept in metadataJson.

Source: Wikipedia, List of quantum processors, CC BY-SA 4.0 — https://en.wikipedia.org/wiki/List_of_quantum_processors (https://creativecommons.org/licenses/by-sa/4.0/). Modified by aiqbench: the table was read from the page's wikitext through the MediaWiki API, references and formatting removed, and each row normalised into aiqbench's common board format with the physical qubit count as the score. The page revision each row was read from is recorded on the row.

Source: Wikipedia, Quantum volume, CC BY-SA 4.0 — https://en.wikipedia.org/wiki/Quantum_volume (https://creativecommons.org/licenses/by-sa/4.0/). Modified by aiqbench: the table was read from the page's wikitext through the MediaWiki API, references and formatting removed, and each row normalised into aiqbench's common board format with log2 of the achieved quantum volume as the score. The page revision each row was read from is recorded on the row.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'simplebench_external.csv' table and rescale its 0-1 scores to percent. Underlying results published by SimpleBench — https://simple-bench.com/, transcribed by Epoch AI.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'simpleqa_verified.csv' table and rescale its 0-1 scores to percent. Evaluation run by Epoch AI.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'swe_bench_verified.csv' table and rescale its 0-1 scores to percent. Evaluation run by Epoch AI.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by SWE-bench Verified leaderboard — https://www.swebench.com/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'swe-bench-verified-leaderboard' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'terminalbench_external.csv' table and rescale its 0-1 scores to percent. Underlying results published by Terminal-Bench (Laude Institute) — https://www.tbench.ai/leaderboard, transcribed by Epoch AI.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Terminal-Bench 4.0 (Harbor / Laude Institute) — https://www.tbench.ai/leaderboard, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'terminal-bench-4.0' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by Holistic Agent Leaderboard (Princeton) — https://hal.cs.princeton.edu/, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'hal-usaco' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: Vectara Hallucination Leaderboard (vectara/hallucination-leaderboard) — https://github.com/vectara/hallucination-leaderboard. Licensed under Apache 2.0. Modified by aiqbench: we parse the ranking table embedded in the repo README between the <!-- LEADERBOARD_START --> and <!-- LEADERBOARD_END --> markers, and store factual-consistency-rate (100 − hallucination rate) as the primary score so "higher is better" holds across every board in the app. The raw hallucination rate, answer rate, and average summary length are preserved on each row in metadata.

Source: metriq-data (Unitary Foundation) — https://github.com/unitaryfoundation/metriq-data, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); hardware runs of the metriq-gym "WIT" benchmark (Apache 2.0) as published to Metriq (https://metriq.info). Modified by aiqbench: one row per device, the most recent eligible run, score = the witness expectation value (0–1) of the 7-qubit WIT circuit. Only runs with num_qubits = 7 are ranked. Runs on simulators are not ranked.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by τ²-bench via the Exgentic Open Agent Leaderboard — https://github.com/sierra-research/tau2-bench, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'tau-bench-2_airline' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by τ²-bench via the Exgentic Open Agent Leaderboard — https://github.com/sierra-research/tau2-bench, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'tau-bench-2_retail' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.

Source: Every Eval Ever datastore (EvalEval coalition), https://huggingface.co/datasets/evaleval/EEE_datastore, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Underlying results published by τ²-bench via the Exgentic Open Agent Leaderboard — https://github.com/sierra-research/tau2-bench, converted into the EEE schema by the coalition's adapters. Modified by aiqbench: the 'tau-bench-2_telecom' collection's headline result per model, normalised into aiqbench's common board format, proportions rescaled to percent.