# Qwen3 8B > Source: Benchize (https://benchize.com) — https://benchize.com/models/qwen3-8b. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/models/qwen3-8b. Qwen Canonical URL: https://benchize.com/models/qwen3-8b Last updated: Wed, 21 Oct 2026 18:52:20 GMT Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations. Qwen ## Benchmarks | Benchmark | Score | ± | Rank | Cost | Latency | Source | | | --- | --- | --- | --- | --- | --- | --- | --- | | [PropensityBench](https://benchize.com/benchmarks/scale%3Apropensitybench.md) | 75.2source score | CI 75.1–75.3 | 3 / 5 | — | — | [scale](https://labs.scale.com/leaderboard/propensitybench) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 75.2 \| source_score \| reported \| Higher is better \| ### Configuration Reasoning effort unknown Model configuration Qwen3-8B sourceModelVersion Not reported thinkingEnabled Not reported ### Reported statistics Uncertainty Confidence interval: 75.1–75.3 contaminationWarning Not reported rank 12 reportedConfidenceHalfWidth 0.1 ### Source [Original source](https://labs.scale.com/leaderboard/propensitybench) Source scale Original model label Qwen3-8B Observed 2026-10-21T16:35:35.949Z Evidence SHA-256 `e65bbee55f6a735247cf59c5746c7850ce3413b5c0686e4c5b71a8cd641ec37f` Source locator `/rows/11` | | | | | | | | | [GPQA Diamond](https://benchize.com/benchmarks/gpqa-diamond.md) | 56.76% | SE ±2.93 pp | 98 / 125 | — | — | [epoch ai · GPQA diamond](https://epoch.ai/data/benchmark_data.zip#file=gpqa_diamond.csv) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.5675505050505051 \| score_fraction \| Best score (across scorers) \| Higher is better \| \| `mean_score` \| 0.5675505050505051 \| score_fraction \| mean_score \| Higher is better \| ### Configuration benchmarkVersion GPQA diamond reasoningBudget Not reported Reasoning effort Not reported sourceModelVersion qwen3-8b ### Reported statistics Uncertainty Standard error: ±2.93 pp standardError 0.029327713464985333 ### Source [Original source](https://epoch.ai/data/benchmark_data.zip#file=gpqa_diamond.csv) Source epoch ai Original model label qwen3-8b Source version GPQA diamond Evaluated 2026-08-27T14:21:43.000Z Observed 2026-10-21T16:30:24.199Z Evidence SHA-256 `4b2dd8376cdd249a2c3065603575208fa726c15eabbb1622689a9abc146ba29a` Source locator `/rows/52` | | | | | | | | | [SVGBench Arena](https://benchize.com/benchmarks/svgbench%3Aarena.md) | 1,282 | — | 167 / 191 | — | — | [svgbench](https://svgbench.ai/api/leaderboard?shape=2) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `rating` \| 1282 \| rating \| reported \| Higher is better \| ### Configuration Reasoning effort unknown sourceModelId model_qwen_qwen3_8b ### Reported statistics coverageCount 27 draws 0 judged 25 losses 18 rank 267 ratingDeviation 28 wins 7 ### Source [Original source](https://svgbench.ai/api/leaderboard?shape=2) Source svgbench Original model label Qwen3 8B Observed 2026-10-21T16:32:23.388Z Evidence SHA-256 `5c0a9c869593820d57d2c05f6d416e1b44a740b4835e66c66e9ad8b8edc59e2c` Source locator `/rows/251` | | | | | | | | | [OTIS Mock AIME 2024–2025](https://benchize.com/benchmarks/epoch-ai%3Aotis_mock_aime_2024_2025.md) | 56.11% | SE ±6.45 pp | 86 / 123 | — | — | [epoch ai · OTIS Mock AIME 2024-2025](https://epoch.ai/data/benchmark_data.zip#file=otis_mock_aime_2024_2025.csv) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.5611111111111111 \| score_fraction \| Best score (across scorers) \| Higher is better \| \| `mean_score` \| 0.5611111111111111 \| score_fraction \| mean_score \| Higher is better \| ### Configuration benchmarkVersion OTIS Mock AIME 2024-2025 reasoningBudget Not reported Reasoning effort Not reported sourceModelVersion qwen3-8b ### Reported statistics Uncertainty Standard error: ±6.45 pp standardError 0.0645035213369854 ### Source [Original source](https://epoch.ai/data/benchmark_data.zip#file=otis_mock_aime_2024_2025.csv) Source epoch ai Original model label qwen3-8b Source version OTIS Mock AIME 2024-2025 Evaluated 2026-08-27T14:21:53.000Z Observed 2026-10-21T16:30:24.199Z Evidence SHA-256 `5b411612f91e155dff4b09d09b548b42f7936482ea8b3c00b510bd46e2f6bc2c` Source locator `/rows/59` | | | | | | | | | [Berkeley Function Calling Leaderboard](https://benchize.com/benchmarks/bfcl%3Aoverall.md) | 40.43% | — | 13 / 26 | $63.95 | 54.17s | [Berkeley Function Calling Leaderboard](https://api.github.com/repos/HuanzhiMao/BFCL-Result/contents/2025-12-16/score/data_overall.csv?ref=cadd76a8dc62984e7bf15ecaf358f070bae745cc) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 40.43 \| percent \| reported \| Higher is better \| \| `cost_total` \| 63.95 \| USD \| evaluation_total \| Lower is better \| \| `latency_mean` \| 54.17 \| seconds \| per_call_mean \| Lower is better \| \| `latency_sd` \| 79.9 \| seconds \| per_call_sd \| Lower is better \| \| `non_live_ast_acc` \| 88.56 \| percent \| reported_task_score \| Higher is better \| \| `non_live_simple_ast` \| 75.25 \| percent \| reported_task_score \| Higher is better \| \| `non_live_multiple_ast` \| 95 \| percent \| reported_task_score \| Higher is better \| \| `non_live_parallel_ast` \| 94.5 \| percent \| reported_task_score \| Higher is better \| \| `non_live_parallel_multiple_ast` \| 89.5 \| percent \| reported_task_score \| Higher is better \| \| `live_acc` \| 80.09 \| percent \| reported_task_score \| Higher is better \| \| `live_simple_ast` \| 84.5 \| percent \| reported_task_score \| Higher is better \| \| `live_multiple_ast` \| 78.92 \| percent \| reported_task_score \| Higher is better \| \| `live_parallel_ast` \| 93.75 \| percent \| reported_task_score \| Higher is better \| \| `live_parallel_multiple_ast` \| 75 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_acc` \| 33.38 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_base` \| 41.5 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_miss_func` \| 38.5 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_miss_param` \| 27 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_long_context` \| 26.5 \| percent \| reported_task_score \| Higher is better \| \| `web_search_acc` \| 13.5 \| percent \| reported_task_score \| Higher is better \| \| `web_search_base` \| 19 \| percent \| reported_task_score \| Higher is better \| \| `web_search_no_snippet` \| 8 \| percent \| reported_task_score \| Higher is better \| \| `memory_acc` \| 13.12 \| percent \| reported_task_score \| Higher is better \| \| `memory_kv` \| 3.87 \| percent \| reported_task_score \| Higher is better \| \| `memory_vector` \| 10.32 \| percent \| reported_task_score \| Higher is better \| \| `memory_recursive_summarization` \| 25.16 \| percent \| reported_task_score \| Higher is better \| \| `relevance_detection` \| 75 \| percent \| reported_task_score \| Higher is better \| \| `irrelevance_detection` \| 82.27 \| percent \| reported_task_score \| Higher is better \| \| `latency_p95` \| 194.15 \| seconds \| per_call_p95 \| Lower is better \| ### Configuration mode Prompt sourceRelease 2025-12-16 ### Reported statistics rank 44 ### Source [Original source](https://api.github.com/repos/HuanzhiMao/BFCL-Result/contents/2025-12-16/score/data_overall.csv?ref=cadd76a8dc62984e7bf15ecaf358f070bae745cc) Source Berkeley Function Calling Leaderboard Original model label Qwen3-8B Observed 2026-10-21T16:39:30.957Z Evidence SHA-256 `ede5081a24bc5ac7f9b3b09a52af6bcfcd35aa0f1709239fed979db77a8159a6` Source locator `/rows/44` | | | | | | | | | [ToneBench Scriptwriting](https://benchize.com/benchmarks/tonebench%3Ascripts.md) | 74.5 | Range 33.3–115.5 | 124 / 126 | — | — | [tonebench](https://benchmark.towardsai.com/data/leaderboard.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 74.5 \| rating \| reported \| Higher is better \| \| `overall` \| 47.15 \| percent \| reported \| Higher is better \| \| `anti_slop_numbers` \| 48.15 \| source_score \| rubric \| Higher is better \| \| `continuity_emotion` \| 44.8 \| source_score \| rubric \| Higher is better \| \| `cue_quality` \| 56.88 \| source_score \| rubric \| Higher is better \| \| `hook_strength` \| 62.78 \| source_score \| rubric \| Higher is better \| \| `length_adherence` \| 28.09 \| source_score \| rubric \| Higher is better \| \| `slop` \| 61.96 \| source_score \| rubric \| Higher is better \| \| `slop_score` \| 75.77 \| source_score \| rubric \| Higher is better \| \| `substance_accuracy` \| 51.27 \| source_score \| rubric \| Higher is better \| \| `tone_voice` \| 42.1 \| source_score \| rubric \| Higher is better \| \| `writing_quality` \| 49.73 \| source_score \| rubric \| Higher is better \| \| `youtube_best_practices` \| 40.2 \| source_score \| rubric \| Higher is better \| ### Configuration provider openrouter Reasoning effort unknown scoreScope 10_task_current sourceConfiguration qwen3-8b-or ### Reported statistics Uncertainty Reported interval: 33.3–115.5 interval [33.3,115.5] rank 131 runCount 5 ### Source [Original source](https://benchmark.towardsai.com/data/leaderboard.json) Source tonebench Original model label Qwen3 8B (thinking on) Evaluated 2026-10-08 Observed 2026-10-21T16:32:48.862Z Evidence SHA-256 `4264338337f418d3ded7ace250cad64fe12cacbd4b79a312bfaf0ae9268a01c3` Source locator `/rows/130` | | | | | | | | | [Berkeley Function Calling Leaderboard](https://benchize.com/benchmarks/bfcl%3Aoverall.md) | 42.57% | — | 10 / 26 | $43.32 | 51.36s | [Berkeley Function Calling Leaderboard](https://api.github.com/repos/HuanzhiMao/BFCL-Result/contents/2025-12-16/score/data_overall.csv?ref=cadd76a8dc62984e7bf15ecaf358f070bae745cc) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 42.57 \| percent \| reported \| Higher is better \| \| `cost_total` \| 43.32 \| USD \| evaluation_total \| Lower is better \| \| `latency_mean` \| 51.36 \| seconds \| per_call_mean \| Lower is better \| \| `latency_sd` \| 76.14 \| seconds \| per_call_sd \| Lower is better \| \| `non_live_ast_acc` \| 87.58 \| percent \| reported_task_score \| Higher is better \| \| `non_live_simple_ast` \| 72.83 \| percent \| reported_task_score \| Higher is better \| \| `non_live_multiple_ast` \| 96.5 \| percent \| reported_task_score \| Higher is better \| \| `non_live_parallel_ast` \| 92 \| percent \| reported_task_score \| Higher is better \| \| `non_live_parallel_multiple_ast` \| 89 \| percent \| reported_task_score \| Higher is better \| \| `live_acc` \| 80.53 \| percent \| reported_task_score \| Higher is better \| \| `live_simple_ast` \| 84.5 \| percent \| reported_task_score \| Higher is better \| \| `live_multiple_ast` \| 79.68 \| percent \| reported_task_score \| Higher is better \| \| `live_parallel_ast` \| 75 \| percent \| reported_task_score \| Higher is better \| \| `live_parallel_multiple_ast` \| 79.17 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_acc` \| 41.75 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_base` \| 50.5 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_miss_func` \| 42 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_miss_param` \| 40 \| percent \| reported_task_score \| Higher is better \| \| `multi_turn_long_context` \| 34.5 \| percent \| reported_task_score \| Higher is better \| \| `web_search_acc` \| 12 \| percent \| reported_task_score \| Higher is better \| \| `web_search_base` \| 15 \| percent \| reported_task_score \| Higher is better \| \| `web_search_no_snippet` \| 9 \| percent \| reported_task_score \| Higher is better \| \| `memory_acc` \| 14.62 \| percent \| reported_task_score \| Higher is better \| \| `memory_kv` \| 5.16 \| percent \| reported_task_score \| Higher is better \| \| `memory_vector` \| 7.1 \| percent \| reported_task_score \| Higher is better \| \| `memory_recursive_summarization` \| 31.61 \| percent \| reported_task_score \| Higher is better \| \| `relevance_detection` \| 93.75 \| percent \| reported_task_score \| Higher is better \| \| `irrelevance_detection` \| 79.07 \| percent \| reported_task_score \| Higher is better \| \| `latency_p95` \| 188.98 \| seconds \| per_call_p95 \| Lower is better \| ### Configuration mode FC sourceRelease 2025-12-16 ### Reported statistics rank 39 ### Source [Original source](https://api.github.com/repos/HuanzhiMao/BFCL-Result/contents/2025-12-16/score/data_overall.csv?ref=cadd76a8dc62984e7bf15ecaf358f070bae745cc) Source Berkeley Function Calling Leaderboard Original model label Qwen3-8B Observed 2026-10-21T16:39:30.957Z Evidence SHA-256 `ede5081a24bc5ac7f9b3b09a52af6bcfcd35aa0f1709239fed979db77a8159a6` Source locator `/rows/39` | | | | | | | | | [EQ-Bench 3](https://benchize.com/benchmarks/eq-bench%3Aeqbench3.md) | 865.5elo | CI 836.7–894.3 | 44 / 50 | — | — | [eq bench · eqbench3](https://eqbench.com/eqbench3.js) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo_norm` \| 865.5 \| elo \| overall \| Higher is better \| \| `rubric_0_100` \| 36.8 \| percent \| overall \| Higher is better \| ### Configuration interactiveAdjustments none protocol eqbench3 sourceModelVersion Qwen/Qwen3-8B ### Reported statistics Uncertainty Confidence interval: 836.7–894.3 confidenceIntervalHigh 894.3 confidenceIntervalLow 836.7 ### Source [Original source](https://eqbench.com/eqbench3.js) Source eq bench Original model label Qwen/Qwen3-8B Source version eqbench3 Observed 2026-10-21T16:35:00.608Z Evidence SHA-256 `3672ff093e739f07ff664a2e6013bc38a6977a146e4092b4cef80ba1780542b2` Source locator `/leaderboardDataEQBench3/67` | | | | | | | | --- Cite as: Qwen3 8B — Benchize, https://benchize.com/models/qwen3-8b (retrieved 2026-10-21)