Expired50% confidenceBenchmarkExact time
On the DeepSWE benchmark, the per-task cost across reasoning tiers is $1 at Low, $1.86 at Medium, $3.47 at High, $4.70 at X-High, and $8.39 at Max.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/14/2026
First Seen
Valid until: 10/12/2026(expired)
Sources
Codex高效使用指南：避免GPT-5.6疯狂烧额度的实用策略
Related Claims
UnverifiedOn the DeepSWE benchmark, a single run with Opus 4.8 costs approximately $8, and a single run with GLM 5.2 Max costs approximately $4.73% similarUnverifiedGPT-5.6 Sol achieved a record-high score of 73% on the DeepSWE benchmark Max tier at a cost of $22 per task, while the competing product Fable scored $839 on the corresponding tier.72% similarUnverifiedBased on current mainstream model API pricing, the cost of a single deep code refactoring task may range between $0.5 and $2.71% similarUnverifiedIn the DiD Suite benchmark, the average cost per task at the Max Effort level is as high as $22, while the cost per task at the Low level is only $3.76, a reduction of over 80%.70% similarPartially VerifiedDeepSeek's API input token price is approximately 1/50 to 1/100 that of Claude Opus70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/507955API
curl https://kongchang.com/api/v1/knowledge/claims/507955MCP
get_claim(id=507955)