Understand this story
ARC Prize announced the evaluation results of Anthropic's Claude Fable 5.2 on the ARC-AGI series of benchmarks. On ARC-AGI-3, it scored 81.2% at a cost of $21.9K under the standard test harness, and 97.7% at $18.6K under the provider adapter harness; on ARC-AGI-2 it scored 93.8% at $0.98 per task; and on ARC-AGI-1 it scored 98.5% at $0.36 per task. ARC Prize stated that Fable 5.2 set a new highest score on ARC-AGI-3 under the standard harness, and that its 93.8% on ARC-AGI-2 is the third-highest among models seen so far, behind only GPT-6 Astra's 95.0% and GPT-6.1 Sol's 94.2%. When using the max reasoning setting plus the standard harness, 28 of the 55 semi-private games hit the 12-hour runtime limit, yielding a score of 65.7%, lower than the 81.2% achieved at the xhigh setting.
AI-generated from public reports
Latest development · Report
ARC Prize announces evaluation results for Claude Fable 5.2 on the ARC-AGI seriesDevelopments
1 confirmed developments
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 10
ARC PrizeFirst-party
ARC Prize announces evaluation results for Claude Fable 5.2 on the ARC-AGI series
ARC Prize announced the results of Anthropic's Claude Fable 5.2 on the ARC-AGI series: it scored 93.8% on ARC-AGI-2 at a per-task cost of $0.98, ranking third among models seen so far, below GPT-6 Astra's 95.0% and GPT-6.1 Sol's 94.2%. It scored 98.5% on ARC-AGI-1 at a per-task cost of $0.36; on ARC-AGI-3 it scored 81.2% under the standard harness at a cost of $21.9K, and 97.7% under the provider adapter harness at a cost of $18.6K, setting a new ARC-AGI-3 high under the standard harness. ARC Prize also noted that, under the standard harness, the max reasoning tier hit the 12-hour runtime limit on 28/55 semi-private games, scoring 65.7%, lower than the xhigh tier's 81.2%.
ARC PrizeFirst-partySelected
ARC Prize publishes evaluation results of Claude Fable 5.2 on ARC-AGI
ARC Prize published Anthropic's Claude Fable 5.2's results on the ARC-AGI series of benchmarks: under the ARC-AGI-3 standard harness it scored 81.2% at a cost of $21.9K, and under the provider adapter harness 97.7% at $18.6K; ARC-AGI-2 was 93.8% at $0.98 per task, and ARC-AGI-1 was 98.5% at $0.36 per task. ARC Prize says Fable 5.2 set a new high score on ARC-AGI-3 under the standard harness, and its 93.8% on ARC-AGI-2 is the third highest among models seen so far, behind only GPT-6 Astra's 95.0% and GPT-6.1 Sol's 94.2%. At the max reasoning tier with the standard harness, 28 of the 55 semi-private games hit the 12-hour runtime limit, scoring 65.7%, lower than the 81.2% at the xhigh tier; the released Fable 5.2's tuning at the medium and xhigh tiers differs from the previously evaluated checkpoint, while the low, high, and max tiers are unchanged.
ARC PrizeFirst-partySelected
ARC Prize publishes Claude Fable 5.2's evaluation results on the various versions of ARC-AGI
ARC Prize published the results of Anthropic's Claude Fable 5.2 on ARC-AGI: under the ARC-AGI-3 standard testing framework it scored 81.2% at a cost of $21.9K, ARC-AGI-2 scored 93.8% at $0.98 per task, and ARC-AGI-1 scored 98.5% at $0.36 per task. The 93.8% on ARC-AGI-2 is the third-highest model score it has seen, behind only GPT-6 Astra's 95.0% and GPT-6.1 Sol's 94.2%. At the max reasoning level, 28 of the 55 semi-private games hit the 12-hour runtime limit, scoring 65.7%, lower than the 81.2% at the xhigh level.
ARC PrizeFirst-party
ARC Prize announces verified results for Claude Fable 5.2 on three ARC-AGI benchmarks
ARC Prize announced the verified results for Anthropic's Claude Fable 5.2 on ARC-AGI: under the standard harness for ARC-AGI-3 it scored 81.2% at a cost of $21.9K, while under the provider adapter harness it scored 97.7% at $18.6K, and it set a new high score on ARC-AGI-3 under the standard harness. It scored 93.8% on ARC-AGI-2 at $0.98/task, and 98.5% on ARC-AGI-1 at $0.36/task. ARC Prize also noted that the max reasoning level, under the standard harness, hit the 12-hour runtime limit on 28 of 55 semi-private games, scoring 65.7%, lower than the 81.2% of the xhigh level.
ARC PrizeFirst-party
ARC Prize evaluates Claude Fable 5.2: 81.2% score on the ARC-AGI-3 standard harness
ARC Prize announced Anthropic Claude Fable 5.2's results on the ARC-AGI series of benchmarks: ARC-AGI-3 scored 81.2% under the standard harness at a cost of $21.9K, and 97.7% at $18.6K under the provider adapter harness; ARC-AGI-2 scored 93.8% at $0.98 per task, and ARC-AGI-1 scored 98.5% at $0.36 per task. At the max reasoning tier, 28 of 55 semi-private games hit the 12-hour runtime limit, scoring 65.7%, lower than the 81.2% of the xhigh tier; ARC Prize believes the max tier would have scored higher without the time limit. ARC Prize also noted that the released Fable 5.2's tuning on the medium and xhigh reasoning tiers differs from the checkpoint used in earlier evaluations, while the low, high, and max tiers are unchanged.