Unverified50% confidenceBenchmarkExact time
Table 1 of the paper shows that ARPO (a pure-text 8B agent without latent-space mode) scores 38.8 on GAIA and 32.0 on WebWalker, both slightly higher than GPT-5.6 Luna.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/19/2026
First Seen
Valid until: 11/2/2026
Sources
Related Entities
Related Claims
UnverifiedWith a 128K token context window, Argon and GPT-6 are nearly tied on long-context accuracy, at 99.7 and 98.7 respectively.67% similarUnverifiedArgon is tied for first place on CWE Bench at 68% alongside GPT-6 Astra and Grok 4.7.64% similarUnverifiedIn Google's published benchmark table, Argon leads or ties GPT-6 Astra, Claude Opus 5.5, and Fable 5.1 in 13 out of 19 evaluation items.63% similarUnverifiedGoogle's methodology PDF states that Argon's Deep SWE result was self-calculated, GPT-6 Astra's figure comes from the DataCurve public leaderboard, and Claude's figure comes from Anthropic's system card.58% similarUnverifiedArtificial Analysis measured that Gemini 4 Argon generates approximately 62,000 output tokens per task, while GPT-6 Astra uses only approximately 27,000 to complete the same work.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1069594API
curl https://kongchang.com/api/v1/knowledge/claims/1069594MCP
get_claim(id=1069594)