# GPT-6 AstraLow > Source: Benchize (https://benchize.com) — https://benchize.com/models/gpt-6-astra-low. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/models/gpt-6-astra-low. OpenAI Canonical URL: https://benchize.com/models/gpt-6-astra-low Last updated: Thu, 22 Oct 2026 06:30:09 GMT Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations. OpenAI ## Benchmarks | Benchmark | Score | ± | Rank | Cost | Latency | Source | | | --- | --- | --- | --- | --- | --- | --- | --- | | [FrontierCode 1.1 extended](https://benchize.com/benchmarks/frontiercode%3Av1_1%3Aextended.md) | 57.4% | — | 48 / 125 | $1.5 | — | [frontiercode](https://cognition.com/data/frontiercode-leaderboard/data.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.574 \| score_fraction \| blocking_criteria_adjusted \| Higher is better \| \| `correct` \| 0.6293 \| score_fraction \| raw_rubric \| Higher is better \| \| `cost` \| 1.4997 \| USD \| per_rollout \| Lower is better \| ### Configuration harness codex Reasoning effort low sourceVersion v1_1 subset extended ### Reported statistics flaggedRate 0 sampleSize 150 ### Source [Original source](https://cognition.com/data/frontiercode-leaderboard/data.json) Source frontiercode Original model label GPT-6 Astra Observed 2026-10-22T05:40:03.220Z Evidence SHA-256 `ff0040778b7bdf69c705ad7474f771fb5e53cfaf5cf8573b3a4614153543d75f` Source locator `/v1_1/data/GPT-6%20Astra/low/extended` | | | | | | | | | [ArXivMath 05/2026](https://benchize.com/benchmarks/matharena%3Aarxiv--may.md) | 88.75% | 95% CI 81.83–95.67% | 4 / 23 | 0.08usd per problem | — | [matharena · arxiv--may](https://matharena.ai/competition_tables/arxiv--may) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 88.75 \| percent \| competition \| Higher is better \| \| `cost_per_problem` \| 0.077 \| usd_per_problem \| competition \| Lower is better \| ### Configuration competition arxiv--may predicted false Reasoning effort low sourceModelName GPT-6 Astra (low) sourceModelVersion openai/gpt-6-astra-low thinkingEnabled true ### Reported statistics Uncertainty 95% confidence interval: 81.83–95.67% confidenceInterval95 6.92 problems 40 ### Source [Original source](https://matharena.ai/competition_tables/arxiv--may) Source matharena Original model label GPT-6 Astra (low) Source version arxiv--may Observed 2026-10-22T05:30:08.858Z Evidence SHA-256 `758e1bba77e36f4ced2a51b4bc0d46bd956ed94ef6f57e93e1a974b75e76a4ce` Source locator `/table/rows/3` | | | | | | | | | [FrontierMath Tier 4 v2 (private)](https://benchize.com/benchmarks/epoch-ai%3Afrontiermath_tier_4_v2.md) | 87.8% | SE ±5.17 pp | 9 / 39 | — | — | [epoch ai · FrontierMath-Tier-4-v2-Private](https://epoch.ai/data/benchmark_data.zip#file=frontiermath_tier_4_v2.csv) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.8780487804878049 \| score_fraction \| Best score (across scorers) \| Higher is better \| \| `mean_score` \| 0.8780487804878049 \| score_fraction \| mean_score \| Higher is better \| ### Configuration benchmarkVersion FrontierMath-Tier-4-v2-Private reasoningBudget Not reported Reasoning effort low sourceModelVersion gpt-6-astra_low ### Reported statistics Uncertainty Standard error: ±5.17 pp standardError 0.051739520574625435 ### Source [Original source](https://epoch.ai/data/benchmark_data.zip#file=frontiermath_tier_4_v2.csv) Source epoch ai Original model label gpt-6-astra Source version FrontierMath-Tier-4-v2-Private Evaluated 2026-08-30T02:22:12.000Z Observed 2026-10-22T03:00:05.281Z Evidence SHA-256 `2bbef8e82ac84071e3ce42bf744f4fc39dff04a549ccb1a288e3800918bc2e55` Source locator `/rows/19` | | | | | | | | | [ARC-AGI-2](https://benchize.com/benchmarks/arc-prize%3Av2_Semi_Private.md) | 85.42% | — | 29 / 243 | 0.42usd per task | — | [arc prize · v2](https://arcprize.org/media/data/leaderboard/v2.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.8541666666666666 \| score_fraction \| v2_Semi_Private \| Higher is better \| \| `cost_per_task` \| 0.4162418333333334 \| usd_per_task \| v2_Semi_Private \| Lower is better \| ### Configuration evaluationSplit v2_Semi_Private harness Not reported modelGroup openai-gpt-6-astra modelType CoT Reasoning effort low sourceModelVersion openai-gpt-6-astra-low thinkingEnabled true ### Source [Original source](https://arcprize.org/media/data/leaderboard/v2.json) Source arc prize Original model label GPT-6 Astra (Low) Source version v2 Observed 2026-10-22T05:20:04.306Z Evidence SHA-256 `4351eb696e78113b0459e9bb5f0258318b905efef34f473dcea51ce641b0d94d` Source locator `/evaluations/229` | | | | | | | | | [BrokenArXiv 08/2026](https://benchize.com/benchmarks/matharena%3Aarxiv_false--august.md) | 74.7% | 95% CI 66.65–82.75% | 7 / 14 | 0.63usd per problem | — | [matharena · arxiv_false--august](https://matharena.ai/competition_tables/arxiv_false--august) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 74.70 \| percent \| competition \| Higher is better \| \| `cost_per_problem` \| 0.63 \| usd_per_problem \| competition \| Lower is better \| ### Configuration competition arxiv_false--august predicted false Reasoning effort low sourceModelName GPT-6 Astra (low) sourceModelVersion openai/gpt-6-astra-low thinkingEnabled true ### Reported statistics Uncertainty 95% confidence interval: 66.65–82.75% confidenceInterval95 8.05 problems 56 ### Source [Original source](https://matharena.ai/competition_tables/arxiv_false--august) Source matharena Original model label GPT-6 Astra (low) Source version arxiv_false--august Observed 2026-10-22T05:30:06.837Z Evidence SHA-256 `1facf024f36359a3fcf069fc3b747d7fbf6aeec67c7949ed674b50ec0447a209` Source locator `/table/rows/6` | | | | | | | | | [BullshitBench v2](https://benchize.com/benchmarks/bullshitbench%3Av2.md) | 64% | — | 37 / 191 | — | — | [bullshitbench · v2](https://api.github.com/repos/petergpt/bullshit-benchmark/contents/data/v2/latest/leaderboard.csv?ref=3c655234d15967738f95da696e3f938d4862dcf7) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `green_rate` \| 0.64 \| score_fraction \| all_attempts \| Higher is better \| \| `red_rate` \| 0.04 \| score_fraction \| all_attempts \| Lower is better \| \| `refusal_rate` \| 0.0 \| score_fraction \| all_attempts \| Lower is better \| \| `avg_score` \| 1.5433 \| score \| judge_grade_0_to_2 \| Higher is better \| ### Configuration Reasoning effort low suite v2 thinkingEnabled true ### Reported statistics acceptedCount 4 answeredCount 100 clearCount 64 errorCount 0 partialCount 32 rank 38 refusalCount 0 sampleSize 100 ### Source [Original source](https://api.github.com/repos/petergpt/bullshit-benchmark/contents/data/v2/latest/leaderboard.csv?ref=3c655234d15967738f95da696e3f938d4862dcf7) Source bullshitbench Original model label openai/gpt-6-astra Source version v2 Observed 2026-10-22T03:10:12.603Z Evidence SHA-256 `98b485e08764d1c97144bc4f7b28ced1a96d50850947c775870dfda0e5d749b5` Source locator `/rows/37` | | | | | | | | | [ARC-AGI-1](https://benchize.com/benchmarks/arc-prize%3Av1_Semi_Private.md) | 96.5% | — | 23 / 245 | 0.23usd per task | — | [arc prize · v1](https://arcprize.org/media/data/leaderboard/v1.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.965 \| score_fraction \| v1_Semi_Private \| Higher is better \| \| `cost_per_task` \| 0.22650770000000006 \| usd_per_task \| v1_Semi_Private \| Lower is better \| ### Configuration evaluationSplit v1_Semi_Private harness Not reported modelGroup openai-gpt-6-astra modelType CoT Reasoning effort low sourceModelVersion openai-gpt-6-astra-low thinkingEnabled true ### Source [Original source](https://arcprize.org/media/data/leaderboard/v1.json) Source arc prize Original model label GPT-6 Astra (Low) Source version v1 Observed 2026-10-22T05:20:04.066Z Evidence SHA-256 `2bf68d22dafba37047d07ad833ebc0e8199c9aa2f9eeaed50c75c1b704462836` Source locator `/evaluations/228` | | | | | | | | | [AutomationBench](https://benchize.com/benchmarks/zapier%3Aautomationbench.md) | 30.29% | — | 17 / 113 | $1.08 | — | [zapier](https://framerusercontent.com/sites/4WTSl4BNjd1q9QFEFibC6h/lOvLnhItj8S8-1vyjH_s33Jn-WmUXPmiJ5Diajl0Q38.CMutaHYl.mjs) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 30.29 \| percent \| reported \| Higher is better \| \| `cost` \| 1.08 \| USD \| per_task \| Lower is better \| ### Configuration benchmarkVersion 1.0.6 costFootnote false Fallback policy Not reported Reasoning effort low Model configuration GPT-6 Astra (Low) thinkingEnabled true ### Reported statistics rank 17 ### Source [Original source](https://framerusercontent.com/sites/4WTSl4BNjd1q9QFEFibC6h/lOvLnhItj8S8-1vyjH_s33Jn-WmUXPmiJ5Diajl0Q38.CMutaHYl.mjs) Source zapier Original model label GPT-6 Astra (Low) Observed 2026-10-21T16:34:56.783Z Evidence SHA-256 `43978e38f018789fa59c11de5328abe9e5905bfa9a5a4d601fa3aa4d9c153b0d` Source locator `/rows/16` | | | | | | | | | [BrokenArXiv 06/2026](https://benchize.com/benchmarks/matharena%3Aarxiv_false--june.md) | 97.69% | 95% CI 94.85–100.53% | 4 / 26 | 0.08usd per problem | — | [matharena · arxiv_false--june](https://matharena.ai/competition_tables/arxiv_false--june) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 97.69 \| percent \| competition \| Higher is better \| \| `cost_per_problem` \| 0.079 \| usd_per_problem \| competition \| Lower is better \| ### Configuration competition arxiv_false--june predicted false Reasoning effort low sourceModelName GPT-6 Astra (low) sourceModelVersion openai/gpt-6-astra-low thinkingEnabled true ### Reported statistics Uncertainty 95% confidence interval: 94.85–100.53% confidenceInterval95 2.84 problems 54 ### Source [Original source](https://matharena.ai/competition_tables/arxiv_false--june) Source matharena Original model label GPT-6 Astra (low) Source version arxiv_false--june Observed 2026-10-22T05:30:06.636Z Evidence SHA-256 `f6d7e2280d05706643d46ae7786bd3fad6d912e4893781c3f62c73deebf38459` Source locator `/table/rows/3` | | | | | | | | | [ArXivMath 06/2026](https://benchize.com/benchmarks/matharena%3Aarxiv--june.md) | 84.38% | 95% CI 77.12–91.64% | 7 / 27 | 0.09usd per problem | — | [matharena · arxiv--june](https://matharena.ai/competition_tables/arxiv--june) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 84.38 \| percent \| competition \| Higher is better \| \| `cost_per_problem` \| 0.088 \| usd_per_problem \| competition \| Lower is better \| ### Configuration competition arxiv--june predicted false Reasoning effort low sourceModelName GPT-6 Astra (low) sourceModelVersion openai/gpt-6-astra-low thinkingEnabled true ### Reported statistics Uncertainty 95% confidence interval: 77.12–91.64% confidenceInterval95 7.26 problems 48 ### Source [Original source](https://matharena.ai/competition_tables/arxiv--june) Source matharena Original model label GPT-6 Astra (low) Source version arxiv--june Observed 2026-10-22T05:30:09.097Z Evidence SHA-256 `6832faff9be7e16a6c646c59017e30a2caedebe8cf02caf617a5ed9d8fb03bab` Source locator `/table/rows/6` | | | | | | | | | [BrokenArXiv 05/2026](https://benchize.com/benchmarks/matharena%3Aarxiv_false--may.md) | 88.5% | 95% CI 82.25–94.75% | 4 / 22 | 0.08usd per problem | — | [matharena · arxiv_false--may](https://matharena.ai/competition_tables/arxiv_false--may) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 88.50 \| percent \| competition \| Higher is better \| \| `cost_per_problem` \| 0.082 \| usd_per_problem \| competition \| Lower is better \| ### Configuration competition arxiv_false--may predicted false Reasoning effort low sourceModelName GPT-6 Astra (low) sourceModelVersion openai/gpt-6-astra-low thinkingEnabled true ### Reported statistics Uncertainty 95% confidence interval: 82.25–94.75% confidenceInterval95 6.25 problems 50 ### Source [Original source](https://matharena.ai/competition_tables/arxiv_false--may) Source matharena Original model label GPT-6 Astra (low) Source version arxiv_false--may Observed 2026-10-22T05:30:06.354Z Evidence SHA-256 `b317fca30109e8c78ed5d2b835fe227b8cfbee3b6f9cb6d7aaeadca866660969` Source locator `/table/rows/3` | | | | | | | | | [FrontierCode 1.1 main](https://benchize.com/benchmarks/frontiercode%3Av1_1%3Amain.md) | 45.27% | — | 48 / 125 | $1.7 | — | [frontiercode](https://cognition.com/data/frontiercode-leaderboard/data.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.4527 \| score_fraction \| blocking_criteria_adjusted \| Higher is better \| \| `correct` \| 0.5 \| score_fraction \| raw_rubric \| Higher is better \| \| `cost` \| 1.7 \| USD \| per_rollout \| Lower is better \| ### Configuration harness codex Reasoning effort low sourceVersion v1_1 subset main ### Reported statistics flaggedRate 0 sampleSize 100 ### Source [Original source](https://cognition.com/data/frontiercode-leaderboard/data.json) Source frontiercode Original model label GPT-6 Astra Observed 2026-10-22T05:40:03.220Z Evidence SHA-256 `ff0040778b7bdf69c705ad7474f771fb5e53cfaf5cf8573b3a4614153543d75f` Source locator `/v1_1/data/GPT-6%20Astra/low/main` | | | | | | | | | [Terminal-Bench 4](https://benchize.com/benchmarks/harbor%3Aterminal-bench-4.md) | 50.61% | 95% CI 47.86–53.36% | 6 / 17 | — | — | [harbor](https://www.tbench.ai/) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 50.61 \| percent \| reported \| Higher is better \| \| `pass_at_2` \| 0.5712 \| score_fraction \| reported \| Higher is better \| \| `pass_at_3` \| 0.6015 \| score_fraction \| reported \| Higher is better \| \| `pass_at_4` \| 0.6212 \| score_fraction \| reported \| Higher is better \| \| `pass_at_5` \| 0.6364 \| score_fraction \| reported \| Higher is better \| ### Configuration agent Codex Reasoning effort low sourceSubmission b3ad58f3-b311-4d4d-875b-3158cf0309d6 ### Reported statistics Uncertainty 95% confidence interval: 47.86–53.36% confidenceHalfWidth95 2.75 sampleSize 330 ### Source [Original source](https://www.tbench.ai/) Source harbor Original model label GPT-6 Astra Evaluated 2026-09-03 Observed 2026-10-22T04:00:05.388Z Evidence SHA-256 `f02bf923eb5a5a2fae5b1e9f0a0c9ff5fe0e3779f0324a9a4f0e9ed6e322db82` Source locator `/rows/13` | | | | | | | | | [DeepSWE](https://benchize.com/benchmarks/deepswe%3Av1.1.md) | 67.04% | Range 65.73–68.34% | 22 / 70 | $1.6 | 613.42s | [deepswe](https://deepswe.datacurve.ai/artifacts/v1.1/leaderboard-live.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_at_1` \| 0.6703539823008849 \| score_fraction \| scored_rollout_attempts \| Higher is better \| \| `pass_at_4` \| 0.7964601769911505 \| score_fraction \| tasks_with_one_passing_rollout \| Higher is better \| \| `cost_per_attempt` \| 1.595223439159292 \| USD \| scored_attempt \| Lower is better \| \| `duration` \| 613.4247787610619 \| seconds \| scored_attempt \| Lower is better \| ### Configuration harness mini-swe-agent Reasoning effort low sourceConfiguration mini_swe_agent_gpt_6_astra_low ### Reported statistics Uncertainty Reported interval: 65.73–68.34% interval [0.6573453717840262,0.6833625928177437] intervalMethod 95% run-to-run: SE across repeated whole-benchmark passes (1.96 * std(runs)/sqrt(R)) runCount 4 sampleSize 452 taskCount 113 ### Source [Original source](https://deepswe.datacurve.ai/artifacts/v1.1/leaderboard-live.json) Source deepswe Original model label gpt-6-astra Observed 2026-10-22T03:50:05.889Z Evidence SHA-256 `a7c15d66288fd249c020b9931c017b92d1a3b90e480b3ff34974b752bd030019` Source locator `/rows/22` | | | | | | | | | [ARC-AGI-3](https://benchize.com/benchmarks/arc-prize%3Av3_Semi_Private.md) | 17.45% | — | 13 / 65 | 38,166.47usd | — | [arc prize · v3](https://arcprize.org/media/data/leaderboard/v3.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.17452190401133805 \| score_fraction \| v3_Semi_Private \| Higher is better \| \| `total_cost` \| 38166.47030000001 \| usd \| v3_Semi_Private \| Lower is better \| ### Configuration evaluationSplit v3_Semi_Private harness standard modelGroup openai-gpt-6-astra modelType CoT Reasoning effort low sourceModelVersion openai-gpt-6-astra-low thinkingEnabled true ### Source [Original source](https://arcprize.org/media/data/leaderboard/v3.json) Source arc prize Original model label GPT-6 Astra (Low) Source version v3 Observed 2026-10-22T05:20:04.578Z Evidence SHA-256 `3ce335487296bb2dcf4ba9546c5a2a02b1907db5b08887cd9ff4429fed824916` Source locator `/evaluations/31` | | | | | | | | | [ArXivMath 08/2026](https://benchize.com/benchmarks/matharena%3Aarxiv--august.md) | 85.09% | 95% CI 78.55–91.63% | 6 / 14 | 1.22usd per problem | — | [matharena · arxiv--august](https://matharena.ai/competition_tables/arxiv--august) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 85.09 \| percent \| competition \| Higher is better \| \| `cost_per_problem` \| 1.22 \| usd_per_problem \| competition \| Lower is better \| ### Configuration competition arxiv--august predicted false Reasoning effort low sourceModelName GPT-6 Astra (low) sourceModelVersion openai/gpt-6-astra-low thinkingEnabled true ### Reported statistics Uncertainty 95% confidence interval: 78.55–91.63% confidenceInterval95 6.54 problems 57 ### Source [Original source](https://matharena.ai/competition_tables/arxiv--august) Source matharena Original model label GPT-6 Astra (low) Source version arxiv--august Observed 2026-10-22T05:30:09.342Z Evidence SHA-256 `5924e5ecc15ff087ec1504b1e4ac44408b005c4f70e81e0d2184f930383a963c` Source locator `/table/rows/5` | | | | | | | | --- Cite as: GPT-6 AstraLow — Benchize, https://benchize.com/models/gpt-6-astra-low (retrieved 2026-10-22)