Unverified50% confidenceFactExact time
On Amazon Bedrock, Nova 2 calls using the portable reasoning: 'xhigh' will now send maxReasoningEffort: 'high' instead of the unsupported 'max', and will return a compatibility warning
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/6/2026
First Seen
Valid until: 1/4/2027
Sources
Related Entities
Related Claims
UnverifiedAccording to discussions in the LinuxDO community, the performance of the Max version may be several tiers higher than that of the xHigh version.55% similarUnverifiedIn the GPQA-Diamond xhigh test, the base model Qwen3.8-27B achieves 88.4% accuracy consuming 6,642 median tokens, while Swift maintains 88.3% accuracy with tokens reduced to 2,77153% similarUnverifiedThe Sol configuration answered all of the first 6 out of 10 runs correctly in HIGH mode, and maintained a perfect score across multiple runs in XH mode.53% similarUnverifiedA post on Reddit claimed that the model GPT-6.1 Sol (max) achieved a score of 100% on the FrontierMath Tier 4 benchmark.52% similarVerifiedOn the DeepSuite benchmark, Grok 4.7 scored 71%, GPT 5.6 Sol scored 72.7%, Fable 5.1 scored 70%, and Astra Max topped the leaderboard at 74.1%.51% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/982582API
curl https://kongchang.com/api/v1/knowledge/claims/982582MCP
get_claim(id=982582)